ROLE OF THE MITOCHONDRIAL PERMEABILITY TRANSITION IN BONE METABOLISM AND AGING

Abstract

The mitochondrial permeability transition pore (MPTP) and its positive regulator, cyclophilin D (CypD), play important pathophysiological roles in aging. In bone tissue, higher CypD expression and pore activity are found in aging, however a causal relationship between CypD/MPTP and bone degeneration needs to be established. We previously reported that CypD expression and MPTP activity are downregulated during osteoblast (OB) differentiation and that manipulations in CypD expression affect OB differentiation and function. Using a newly developed OB-specific CypD/MPTP gain-of-function (GOF) mouse model, we here present evidence that overexpression of a constitutively active K166Q mutant of CypD (caCypD) impairs OB energy metabolism and function, and bone morphological and biomechanical parameters. Specifically, in a spatial-dependent and sex-dependent manner, OB-specific CypD GOF led to a decrease in oxidative phosphorylation (OxPhos) levels, higher oxidative stress, and general metabolic adaptations coincident with the decreased bone organic matrix content in long bones. Interestingly, accelerated bone degeneration was present in vertebral bones regardless of sex. Overall, our work confirms CypD/MPTP overactivation as an important pathophysiological mechanism leading to bone degeneration and fragility in aging.

INTRODUCTION:

Bone maintenance is a dynamic process aimed to balance the activity of bone-forming and bone-resorbing cells in order to sustain a healthy bone mass. However, as we age, bone resorption tends to gradually outpace bone formation, which can cause increased bone loss and osteoporosis. Characterized by impaired osteoblast (OB) activity, aged-bone phenotype presents with decreased bone structural and biomechanical parameters^(1–3). Several groups including ours have shown that OB differentiation and bone matrix synthetic function require high levels of mitochondrial oxidative phosphorylation (OxPhos)^(4–8), We and others also reported that bone degeneration in aging is associated with mitochondrial dysfunction and oxidative stress in OBs and osteocytes (OT) ^{(2, 9, 10)}. Higher levels of oxidative stress observed in aging present a challenging environment for cell and mitochondrial function feeding a ‘vicious circle’ involving mitochondrial DNA (mtDNA) damage, further generation of reactive oxygen species (ROS), decreased expression of respiratory chain proteins, deficient mitophagy, and dysfunctional mitochondrial stress response^(11–14). These alterations create extensive changes in mitochondrial function and morphology, ultimately affecting overall cell energy metabolism and prompting metabolic adaptions that favor catabolic pathways.

The mitochondrial permeability transition pore (MPTP) is a central mechanism controlling directly and indirectly mitochondrial response to pathological stresses. In physiological conditions, higher MPTP opening activity is associated with decreased electron flow into the electron transport chain (ETC) and compensatory activation of aerobic glycolysis. Encoded by the nuclear gene PPIF, Cyclophilin D (CypD) is imported into mitochondria where it performs several functions including positive regulation of MPTP opening. Under certain conditions, such as calcium overload, CypD binds to ATP synthase, likely triggering its oligomerization, and thereby forming the MPTP^(15–18). MPTP opening leads to the loss of integrity of the inner mitochondrial membrane (IMM). In several studied tissues, OxPhos activation during cell differentiation requires CypD downregulation and MPTP closure to establish the mitochondrial membrane potential (Δψm), setting the proton motive force (Δp) and coupling respiratory chain to ATP production^(19–21). In fact, we recently demonstrated that during OB differentiation, MPTP is progressively inhibited due to transcriptional of PPIF and, therefore, expression of CypD⁽²²⁾. In aging, higher CypD expression and pore activity are described in several tissues and associated with all above-mentioned mitochondrial and cell energy metabolic changes^{(13, 23, 24)}. Not surprisingly, we reported higher levels of CypD and MPTP activity in aged mouse bones when compared to young mice⁽²²⁾.

CypD manipulation is vastly studied using global knockout mouse models. Its overexpression or enhanced activity is associated with disease onset and progression, whereas its downregulation or inhibition is protective against aging and some degenerative pathologies. In bone, global genetic deletion of Ppif was shown by us to improve fracture healing and protect against aged-bone phenotype^{(10, 22, 25)}. Conversely, caCypD expression in OBs of OB-specific CypD conditional knockout (cKO) recovers aged-bone phenotype, characterized by decreased morphological and biomechanical parameters⁽²²⁾. Causal relationship between increased CypD expression/MPTP activity and bone degeneration needs to be further established. Herein, we took advantage of a recently created by us tissue-specific caCypD expression, (R26^caPpif) mouse model to target Cre^Col1-expressing OBs to explore how CypD/MPTP GOF contributes to bone aging.

RESULTS

CypD/MPTP gain-of-function decreases osteoblast function and mitochondrial OxPhos parameters in vitro

To establish a mechanistic link between high MPTP activity and decreased OB function, the phenomena reported by us before in aged mouse bones, we utilized an OB-specific tamoxifen-inducible CypD GOF mouse model, 2.3kb Col1^CreERt2/+;R26^{caPpif/+ (22)}. This mouse model expresses constitutively active K166Q (acetylation-mimetic) mutant of Ppif (caPpif) followed by IRES-eGFP sequence. To confirm our previously published results⁽²²⁾, we isolated bone marrow stromal cells (BMSCs) from these skeletally mature (3 months old) CypD GOF male mice. Cells were expanded and osteoinduced in osteogenic media, and Cre-mediated recombination was initiated by 4’-OH-Tamoxifen supplementation. Control Cre⁻ cells were also treated with 4’-OH-Tamoxifen to account for its possible effects on osteogenic cells. We have previously described that these CypD GOF cells show increased MPTP opening activity⁽²²⁾. Cells were collected at day 14 of osteoinduction and analyzed. Cre⁺ and GFP⁺, i.e. caCypD-expressing, OBs (Fig. 1A, images in the left column) presented with impaired osteogenic differentiation. This was manifested by decreased alkaline phosphatase (ALP) staining, mineralization capacity measured with Alizarin Red (AR), collagen formation measured with Sirius Red (SR), and OB genetic markers, Bglap and Col1a1, when compared to Cre⁻ GFP⁻ control cells (Figure 1A, B). CaCypD expression and higher MPTP activity are known to impair mitochondrial OxPhos and shift cell metabolism towards glycolysis⁽²⁶⁾. To confirm these phenomena, we used the XF96 Seahorse Bioanalyzer and evaluated the metabolic profile of CypD GOF OBs compared to Cre⁻ control cells. Using the differentiation and recombination protocol described above, we measured cellular oxygen consumption rate (OCR) as a measure of OxPhos and TCA cycle activity, and extracellular acidification rate (ECAR) as a measure of glycolysis. All OxPhos parameters were decreased in CypD GOF OBs with the exception of proton leak that showed no difference between groups (Figure 1C–D and Supplementary Figure S1A). Conversely, glycolytic basal rate is increased in CypD GOF OBs (Figure 1E), suggesting a glycolytic shift in the presence of a higher MPTP activity. MtDNA expression was slightly higher in CypD GOF OBs (Supplementary Figure S1B), likely as a compensatory effect for the impaired function^{(10, 27)}.

Figure 1. CypD gain-of-function impairs osteogenic differentiation, collagen biosynthesis, and mitochondrial function and network in osteoblasts.

BMSCs from Col1^CreERt2;R26^caPpif/+ or R26^caPpif/+ control mice were cultured in osteogenic media for 11 to 14 days. A) Staining representative image at day 14 for eGFP signal, Alkaline Phosphatase (ALP), Alizarin Red (AR), and Sirius Red (SR): CypD GOF (tamoxifen-induced Cre⁺) BMSCs showed decreased OB activity, mineralization capacity, and collagen formation; B) Real-time RT-PCR data: Col1^CreERt2;R26^caPpif/+ BMSCs showed decreased OB differentiation markers; C) Seahorse traces showing oxygen consumption rate (OCR) for BMSCs from Col1^CreERt2;R26^caPpif/+ mice osteoinduced for 14 days presented decreased OxPhos parameters when compared to Cre⁻ control cells. Arrows indicate time point of injection of metabolic inhibitors: (carbonyl cyanide p-trifluoromethoxyphenylhydrazone: FCCP, antimycin A: AA, rotenone: ROT); D) Specific metabolic contributions to OCR were calculated as Basal Respiration (baseline minus AA/Rot), ATP production (baseline minus Oligomycin), Maximum Respiration (FCCP minus AA/Rot) Spare Capacity (FCCP minus baseline), and Proton Leak (Oligo minus AA/Rot); E) Specific metabolic contributions to extracellular acidification rate (ECAR) calculated as basal ECAR (baseline minus 2-DG) showed increased basal rate in Cre⁺ cells; F) and G) Representative image from R26^caPpif/+ BMSCs infected with Adeno-Cre virus and cultured in osteogenic media for 11 days; F) Osteoinduced R26^caPpif/+ BMSCs infected with Adeno-Cre showed no difference in ROS (MitoSox) signal when compared to Cre⁻ cells. However, 1mM H₂O₂ addition led to an increased response of ROS production in Cre⁺ cells; G) Osteoinduced R26^caPpif/+ BMSCs infected with Adeno-Cre and stained with CMXRos showed decreased mitochondrial morphology and network parameters for Cre⁺ cells, number of mitochondrial endpoints per cell normalized to total mitochondrial area is increased in Cre⁺ cells. Mitochondria are identified in imageJ using a convolution filter and thresholding for intensity and minimum pixel area. Particle analysis detects and calculates measurements for individual mitochondrion. Plot shows the actual data points (biological replicates), calculated means and P value vs Cre⁻ controls determined by an unpaired t-test. For Seahorse assay, n=15 (three biologic replicates of five technical replicates). For mitochondrial network analysis, n=9, a total of 6578 mitochondria were analyzed in the Cre⁻ group and 2499 mitochondria were analyzed in the Cre⁺ group.

To further evaluate mitochondrial function and dynamics in the CypD GOF model, we collected BMSC from R26^caPpif/+ mice and induced recombination through adeno-Cre viral infection. The advantage here is that we can monitor GFP⁺ CypD GOF cells and GFP⁻ control cells in the same field of view as only a portion of cells get infected. Transduced BMSCs were then induced in osteogenic media for 11 days, stained with the nuclear fluorescent stain, Hoechst33342, and various mitochondrial fluorescent probes. First, we measured mitochondrial ROS levels with the mitochondrial ROS-specific probe, MitoSox, and found that CypD GOF did not change MitoSox signal under basal conditions. However, when OBs were challenged with H₂O₂, CypD GOF cells showed a steep increase in the mitochondrial ROS production, when compared to controls (Figure 1F). Next, we calculated several mitochondrial networking parameters using MitoTracker Red CMXRos probe. When compared to controls, CypD GOF OBs showed significant changes in all morphological parameters indicating decreased mitochondrial fusion and poor network formation (Figure 1G and Supplementary Figure S1C–G), with the exception of mitochondrial relative elongation which showed no difference between the groups (Supplementary Figure S1H). Interestingly, undifferentiated BMSCs displayed no differences in mitochondrial morphological and networking parameters between CypD GOF and control cells (Supplementary Figure S2A–H). This is likely because mitochondria in undifferentiated BMSCs present as more punctuated, smaller and separate organelles and not as a network, consistent with previous reports from our group and others ^{(5, 28)}.

Altogether, these results indicate that CypD/MPTP gain-of-function disrupts OxPhos, increases oxidative stress in OBs, and impairs OB differentiation and bone matrix synthesis in vitro.

Osteoblast-specific CypD/MPTP gain-of-function decreases osteoblast function in vivo

Next, we sought to determine how OB-specific CypD/MPTP GOF affects OB function in vivo. We again used the OB-specific CypD GOF mouse (2.3kb Col1^CreERt2/+;R26^caPpif/+) vs Cre⁻ control (R26^caPpif/+) described above. OB-specific Cre-mediated recombination and, thus, caPpif overexpression was induced beginning at 2 months of age. Tissue was harvested at 4Mo or 12Mo while maintaining recombination with bimonthly tamoxifen boosts (Figure 2A). We chose the 12Mo time point because C57BL/6 (our mice genetic background) and other strains loose significant amount but not yet all trabecular bone by 12Mo. Therefore, we considered it the optimal time point to detect possible changes in the rate of bone degeneration with age caused by OB-specific CypD GOF. GFP signal confirming a successful recombination and caCypD expression was present in the bones of Cre⁺ but not Cre⁻ mice. Additionally, no significant differences in transgene expression were found between sex, age, and gender groups (Supplementary Figure S3A). To reassure that GFP signal is a readout of Col1^CreERt2-mediated recombination specificity and therefore, caPpif expression in mature OBs, we harvested BMSCs from 4 Mo-old OB-specific CypD GOF and Cre⁻ control mice (all in triplicates). Cells were expanded and osteoinduced in osteogenic media, and Cre-mediated recombination was initiated by 4’-OH-Tamoxifen supplementation starting at day 0 of osteogenic differentiation. Expression of caCypD mutant was detected in Cre⁺ cells by western blot at day 14 but not at day 7 of osteoinduction (Supplementary Figure S3B). This result confirms our previous report that only mature, Col1-expressing OBs are targeted by 2.3kB Col1^{CreERt2 (22)}.

Figure 2. CypD gain-of-function decreases osteoblast activity in vivo.

A) Experimental design: recombination and therefore, CypD GOF was induced in 2 mo-old Col1^CreERt2/+;R26^caPpif/+ mice. Tissue was harvested at 4Mo, or 12Mo preceded by bimonthly boost injections; B) Serum biomarker for OB activity, P1NP, is decreased in the Cre⁺ mice from 4Mo cohort. Osteocalcin levels are decreased in the vertebrae of females at 4Mo (C) and males at 12Mo (D). E) representative sections and quantitative analyses of mouse bones labeled with Calcein for dynamic Mineral Apposition Rate (MAR), and Bone Formation Rate (BFR) assay. Dynamic histomorphometry analysis revealed no differences in OB activity when Cre⁺ and Cre⁻ control mice were compared in 12Mo females, but 12Mo males showed decreased BFR. Males and females were analyzed separately to account for potential sexual dimorphism. Plots show the actual data points from four to nine independent mice per group. Calculated means and P value vs Cre⁻ control mice were determined by an unpaired t-test.

Several assays were used to measure bone turnover. In the young 4Mo-old cohort, serum P1NP, a biomarker for OB activity in producing collagen type I, showed decreased levels in Cre⁺ mice when compared to Cre⁻ controls (Figure 2B). This result suggests that, consistent with the above in vitro results, OB-specific CypD GOF impairs collagen synthesis early on in vivo. These data also indicate that in young mice, OB-specific CypD GOF impairs deposition of organic matrix by OBs. This was further confirmed by decreased osteocalcin (Ocn) levels detected via immunofluorescence and reaching statistical significance in Cre⁺ females at 4Mo (Figure 2C). At 4Mo, mice did not show any differences in accumulation of infrared BoneTag bone formation probe⁽²⁹⁾ in the axial (caudal vertebrae) or long (hind limb) bones as measured via in vivo IVIS-mediated imaging (Supplementary Figure S3B, C). These results were confirmed with more widely used method of dynamic bone formation using calcein double labeling (Supplementary Figure S3D). Lack of changes in BoneTag and calcein labeling in young Cre⁺ mice may be due to the fact that these techniques mostly reflect deposition of inorganic, calcified matrix which is not yet significantly affected at 4Mo. Bone resorption and osteoclast (OC) function markers, serum CTX-I and bone TRAP staining (Supplementary Figure S3E–G), did not show any differences between studied groups.

In the 12Mo cohort vs 4Mo cohort, P1NP levels were dramatically decreased in Cre⁻ control mice and remained low in Cre⁺ mice, so that the differences between the Cre⁻ and Cre⁺ mice observed at 4Mo were no longer noticeable (Supplementary Figure S4A). Ocn levels were significantly decreased in the vertebrae of Cre⁺ vs Cre⁻ male mice (Figure 2D), suggesting that OB function in vertebral bones is affected early on in females and later in males. BoneTag labeling at 12Mo was significantly decreased in the axial but not long bones in Cre⁺ vs control mice (Supplementary Figure S4B, C). Dynamic histomorphometry of the tibiae revealed no differences in the mineral apposition rate (MAR) between groups in agreement with BoneTag labeling, and a significant decrease in bone formation rate (BFR) only in Cre⁺ males (Figure 2E). No differences in serum CTX-I or vertebral TRAP staining were found (Supplementary Figure S4D, E), however a decreased TRAP staining was observed in long bones of Cre⁺ male mice when compared to controls (Supplementary Figure S4F). Therefore, in 12Mo-old middle-aged mice, OB-specific CypD GOF caused a significant decline in OB function in the axial bones in both sexes and in long bones in males.

In sum, these results indicate that OB-specific CypD/MPTP GOF accelerates aging-associated decline in OB function which is more pronounced in the axial skeleton and in males.

Osteoblast-specific CypD/MPTP gain-of-function leads to accelerated bone aging in the spine in both sexes and in long bones in males

Decreased P1NP serum biomarker levels and OB activity in vertebral bones are described in aging and osteoporotic bone phenotype^{(10, 30)}. In fact, vertebral bones are the first bones being affected by osteoporosis and bone loss detected by dual-energy X-ray absorptiometry (DEXA) scanning in humans⁽³¹⁾. Therefore, we used DEXA scanning to evaluate vertebral and whole-body composition in our mouse model in the 4Mo and 12Mo cohorts. To further investigate our model, we aggregated bone morphological and biomechanical data from all cohorts to analyze 4Mo and 12Mo mice using intra-cohort (Cre⁺ vs Cre⁻) and inter-cohort (12Mo vs 4Mo) comparisons. DEXA scanning revealed no changes in bone mineral density (BMD) when comparing Cre⁺ and Cre⁻ mice in either the whole skeleton (Supplementary Figure S5B) or vertebral L5/L6 bones (Supplementary Figure S5C). However, fat percentage in the whole body and L5/L6 lumbar region was increased in Cre⁺ mice only in females in the 4Mo cohort (Supplementary Figure S5D, E). Since no significant changes in total body mass were detected between experimental and control groups (Supplementary Figure S5A), it suggests a decrease in total body lean mass and increase in bone marrow fat in the spine in Cre⁺ female mice at 4Mo. Further analysis into the vertebral bone morphological structure in the 4Mo cohort using microCT scanning revealed no differences between Cre⁺ mice and Cre⁻ controls (Figure 3A–E and Supplementary Figure S5F–H). However, 12Mo cohort showed that several morphological parameters in Cre⁺ mice were decreased vs Cre⁻ controls. For instance, OB-specific CypD GOF led to significant decrease in trabecular BMD in the spine in both males and females. (Supplementary Figure S5G). On the other hand, only Cre⁺ males showed decreased measurements for cortical BMD and thickness as well as trabecular thickness and bone fractional volume in the spine (Supplementary Figure S5F, G and Figure 3B, D, E). No differences were found in separation between Cre⁺ and Cre⁻ groups at 12Mo (Supplementary Figure S5H). Despite the fact that females did not show significant changes between Cre⁺ and Cre⁻ mice at 12Mo in bone morphological parameters, inter-cohort comparison revealed that OB-specific CypD GOF significantly decreased bone fractional volume in females (Figure 3E). It suggests that CypD overexpression poses a mechanism for accelerated bone aging and degeneration in females as well. In fact, these changes were sufficient to decrease biomechanical properties and increase bone fragility in the vertebrae of both male and female Cre⁺ mice (Figure 3F–G). Twelve month is not an advanced age for many inbred mouse strain including C57BL/6J, our mice genetic background. At 12Mo, bone shows only a moderate decline in biomechanical properties in the spine as obvious from our data shown in Figure 3F–G. When compared to 4Mo, all measured parameters at 12Mo were not significantly changed in Cre⁻ control females, and only two of the three (Yield and Maximal Load) showed significance in Cre⁻ control males. However, in Cre⁺ males and females, all measured parameters were significantly decreased at 12Mo when compared to 4Mo.

Figure 3. Bone volumetric parameters and biomechanical properties are decreased in the vertebrae of mice with osteoblast-specific CypD gain-of-function.

Bone density and volumetric properties of vertebrae from Col1^CreERt2/+;R26^caPpif/+ and control R26^caPpif/+ mice, and biomechanical properties were measured by μCT and a compression test. A) and C) Representative cortical and trabecular vertebral bone images, respectively; B) Cortical thickness, D) Trabecular thickness, and E) Bone fractional volume is decreased only in males in the 12Mo cohort. However, cortical thickness (Ct.Th) and bone over total volume (BV/TV) are decreased in the 12Mo vs 4Mo Cre⁺ female mice; F) Bone stiffness is decreased in Cre⁺ females within the 4Mo cohort and in Cre⁺ males within the 12Mo cohort vs Cre⁻ mice. Both Cre⁺ males and females but not Cre⁻ mice show decreased bone stiffness at 12Mo vs 4Mo; G) Yield load and Maximum load (H) showed similar trend observed for Stiffness. Plots show the actual data points from eight to twelve independent mice per group. Males and females were analyzed separately to account for potential sexual dimorphism. Calculated means and P value determined by an unpaired t-test.

Taken together, these results indicate that activation of CypD/MPTP in OBs is in fact an important driver of bone degeneration in the axial skeleton in both males and females.

Osteoblast-specific CypD/MPTP gain-of-function produced sexually dimorphic morphological effects and decreased unmineralized matrix and proline content in long bones

We have previously shown that caCypD expression in OBs of aged 24Mo-old OB-specific CypD cKO mice decreases long bone morphological parameters and biomechanical properties⁽²²⁾. Additionally, we showed above that bone turnover was different between long and axial bones in response to OB-specific CypD GOF with higher prevalence in males. It suggests that mitochondrial function and, therefore, cell metabolism might affect bone maintenance differently depending on sex and bone type. To further investigate such questions, we analyzed femurs in the same experimental design and conditions as vertebrae presented above. Bone morphological and biomechanical data were aggregated to analyze 4Mo and 12Mo mice using intra-cohort (Cre⁺ vs Cre⁻) and inter-cohort (12Mo vs 4Mo) comparisons. We observed that when compared to Cre⁻ control mice, both cortical and trabecular BMDs were decreased in male and female Cre⁺ mice at 4Mo but not at 12Mo. Interestingly, trabecular BMD was increased in Cre⁺ females at 12Mo (Supplementary Figure S6A, B). Cortical thickness was decreased in Cre⁺ male but not female mice at 12Mo (Figure 4B). Analysis of trabecular bone parameters showed that OB-specific CypD GOF had minor effects on maintenance of the trabecular structure of long bones. No significant intra-cohort differences were observed in trabecular thickness, therefore inter-cohort changes showed in the analysis (Figure 4D) are not due to OB-specific CypD GOF but regular aging process. Variations in trabecular bone fractional volume, trabecular number and space measurements presented similar trend as in trabecular thickness assessment, but some significant inter-cohort differences at 4Mo were noted vs controls: 1) bone fractional volume is increased in Cre⁺ females; 2) trabecular number is decreased in both males and female Cre⁺ mice; and 3) CypD GOF led to an increase in trabecular space in both male and female Cre⁺ mice (Figure 4E–G). Since no significant differences were found inter-cohort at 12Mo, we can conclude that changes seen inter-cohort are also due to regular aging process and are not caused by OB-specific CypD GOF. A possible explanation here is that the regular aging process that was shown by us to increase CypD levels in bone⁽²²⁾, is catching up and matching the long bone phenotype observed in OB-specific CypD GOF mice.

Figure 4. Long bone phenotype showed pronounced sexual dimorphism in response to osteoblast-specific CypD gain-of-function.

Volumetric and biomechanical properties of femurs from Col1^CreERt2/+;R26^caPpif/+ and control R26^caPpif/+ mice were measured by μCT and a three-point bending test, respectively. Aggregated data comparing 4Mo and 12 Mo cohorts: A) and C) Representative 3D-reconstructed μCT images of cortical and trabecular femoral bone; B) Cortical thickness is decreased in males only in Cre⁺ vs Cre⁻ mice at 12Mo; D) Trabecular thickness showed no differences in male mice when compared intra and inter-cohort, but it is increased for females in both Cre⁻ and Cre⁺ inter-cohort; E) Bone fractional volume decreases equally in Cre⁻ and Cre⁺ males and females, when comparing 12Mo vs 4Mo; F) Trabecular number followed similar trend as BV/TV; G) Trabecular space is decreased for both males and females at 4Mo intra-cohort, but increased in all groups inter-cohort; H) Biomechanical 3 point-bending test showed significant decrease in yield load only in Cre⁺ vs Cre⁻ male mice at 12Mo. Plots show the actual data points from eight to twelve independent mice per group. Males and females were analyzed separately to account for sexual dimorphism. Calculated means and P value determined by an unpaired t-test.

To investigate how these combined bone structural changes affect biomechanical properties in OB-specific CypD GOF mice, we performed a three-point bending test in femurs. We found no differences in Young’s modulus and maximum load inter-cohort (Supplementary Figure S6C, D), however yield load was decreased in Cre⁺ males at 12Mo (Figure 4H). Yield load is the maximum stress a structure can support before a permanent deformation is induced, describing the material elastic property⁽³²⁾. The elastic behavior of long bones is mostly determined by unmineralized matrix quality and collagen structure⁽³³⁾. Therefore, we used Raman spectrometry, a non-invasive optical technique that measures chemical composition of bone based on the inelastic scattering of light. The intensity peaks in Raman spectra are related to specific compounds present within the sample; and their amount can be calculated as a mineral/non-mineral ratio⁽³⁴⁾. Humeri from 12Mo-old mice were analyzed for their composition as long bone samples. Principal component analysis (PCA) revealed that differences in the amount of phosphate (PO₄⁻³), carbon content (CH₂), and amide I in bone from males at 12Mo are sufficient to explain the differences in bone composition between Cre⁺ mice and Cre⁻ controls (Figure 5A). However, we found no differences in bone chemical composition in the 4Mo cohort (Supplementary Figure S7A, B) and in females at 12Mo (Figure 6A), matching what we have described for biomechanical testing. Next, we evaluated bone crystallinity and carbonate (CO₃-²)/PO₄⁻³ ratio and since we found no differences between groups (Figure 4B and Supplementary Figure S7E), we confirmed that bone mineral matrix content is not affected by OB-specific CypD GOF in long bone. However, when we evaluated the ratio of proline, amide I, or CH₂ to CO₃-² or PO₄⁻³ (Figure 5C–H) we found increased ratio values confirming that OB-specific CypD GOF decreases organic matrix content and structure in males at 12Mo. As pointed in the PCA analysis, amide III ratios do not change between groups (Supplementary Figure S6F, G).

Figure 5. Organic component of bone matrix is decreased in long bones from osteoblast-specific CypD gain-of-function male mice.

Chemical composition of the humerus from Col1^CreERt2/+;R26^caPpif/+ and control R26^caPpif/+ mice at 12Mo was detected with Raman spectrometry: A) Principal component analysis 1 indicates significant differences between Cre⁺ mice and Cre⁻ controls only in males for phosphate (PO₄⁻³), carbon content (CH₂), and amide I; B) Crystallinity show no differences between experimental and control groups; C) PO₄⁻³/proline, CO₃⁻²/proline (F); PO₄⁻³/amide I (D), CO₃⁻²/amide I (G); PO₄⁻³/CH₂ (E) and CO₃⁻²/CH₂ (H) ratios are increased in Cre⁺ male mice when compared to Cre⁻ controls. Plots show the actual data points from four to five independent mice per group. Males and females were analyzed separately to account for sexual dimorphism. Calculated means and P value vs Cre⁻ control mice was determined by an unpaired t-test.

Figure 6. Significant metabolic changes are detected in bone tissue from osteoblast-specific CypD gain-of-function mice at 4Mo.

Small metabolites were extracted from tibial bone shafts of Col1^CreERt2/+;R26^caPpif/+ and R26^caPpif/+ control mice at 4Mo and analyzed using LC-MS. A) and E) Volcano plots showing all captured metabolites as a function of log(P value) over log(fold change) between Cre⁺ vs Cre⁻ samples. Metabolites are grouped into appropriate metabolic pathways: B) and F) Glycolysis; C) and G) TCA cycle; D) and H) Bioenergetic and RedOx Ratios: energy charge ([ATP + 0.5 ADP] /[ATP + ADP + AMP]), reduced to oxidized glutathione ratio (GSH/GSSG), oxidized to reduced NADP ratio (NADP+/NADPH), oxidized to reduced NAD ratio (NAD⁺/NADH). Plots show the actual data points from seven independent mice per group. Males and females were analyzed separately to account for sexual dimorphism. Calculated means and P value vs Cre⁻ control mice determined by an unpaired t-test. Two and a half fold change over Cre⁻ group was set as significancy level for metabolite changes.

Taken together, these results show that in long bones, OB-specific CypD GOF accelerates bone degeneration during the aging process in males with insignificant effects in females. Bone quality analyses indicate that this is mostly due to the effect on the organic component of bone matrix, consistent with the impaired OB function described above.

Bone metabolome of mice with OB-specific CypD GOF is consistent with mitochondrial dysfunction, oxidative stress, decreased bone matrix anabolism, and premature aging

Mitochondrial dysfunction is regarded as an important hallmark of aging. Generally, function and integrity of mitochondria are impaired during aging prompting metabolic adaptations to cope with decreased OxPhos function^{(11–14, 35)}. Indeed, we have demonstrated before a significant glycolytic shift and alterations in intermediates and metabolite levels in aged mouse bones compared to bones from young mice⁽¹⁰⁾. To explore how CypD/MPTP GOF mimicking CypD overexpression and MPTP activation observed in aging, affects bone metabolome, we performed steady-state liquid chromatography–mass spectrometry (LC-MS) analysis of tibial shafts from our experimental model depicted above. We found that in 4Mo-old males, OB-specific CypD GOF produced minor changes in amino acids, glycolytic and TCA cycle intermediates (Supplementary Figure S8B and Figure 7A–C). The top hit for metabolite enrichment analysis was β-alanine metabolism (Supplementary Figure S8A). Interestingly, β-alanine is described to increase mitochondrial biogenesis⁽³⁶⁾. It is not clear however, whether such changes in β-alanine metabolism are due to a decreased metabolic turnover or a compensatory effect in Cre⁺ male mice. No changes in calculated ratios used to describe metabolic and oxidative stress, such as energy charge, oxidized to reduced glutathione (GSH/GSSG), NADP⁺/NADPH, and NAD⁺/NADH ratios, were found between Cre⁺ vs Cre⁻ male mice at 4Mo (Figure 6D). Conversely, OB-specific CypD GOF in females at 4Mo presented important changes in metabolite levels interconnecting glycolytic, pentose phosphate (PPP), Krebs cycle, and fatty acid synthesis pathways (Figure 6E–G and Supplementary Figure S8B). Not surprisingly, Warburg effect was the top hit in the metabolite enrichment analysis in females (Supplementary Figure S8D). A higher activity of the glycolytic pathway requires a higher demand for glucose. Glycogenolysis is the breakdown of glycogen into glucose-1-phosphate. UDP-glucose comes from the glycogen first breakdown step. Since UDP-glucose would be upregulated in the event of glycogen anabolism, what we are seeing in fact is glycogen catabolism in the gluconeogenesis pathway (Supplementary Figure S8F). Another important observation is that increased malonyl-CoA levels which are known to inhibit mitochondrial fatty acid (FA) uptake and oxidation. Malonyl-CoA is a negative regulator for carnitine, a transporter of long-chain FAs into mitochondria for FA oxidation⁽³⁷⁾. Since acetoacetate upregulation is shown during FA synthesis (Supplementary Figure S8G), we can assume that OB-specific CypD GOF in females is activating FA synthesis in bone tissue. Accordingly, we detected higher fat percentage within bone tissue in Cre⁺ females at 4Mo when DEXA scan was performed. NAD⁺/NADH and NADP⁺/NADPH ratios were found decreased in Cre⁺ females, both indicating the presence of mitochondrial dysfunction consistent with premature aging. Overall, at 4Mo, OB-specific CypD GOF in females produced more alterations in metabolite levels than in males. However, such changes in Cre⁺ female mice were also accompanied by an upregulation in metabolites described to have anti-inflammatory effects ^(38–40) (Supplementary Figure S8H), which was not observed in males at 4Mo (Supplementary Figure S8C). Surprisingly, at 12Mo, very few changes in metabolite levels were observed between Cre⁺ and Cre⁻ female mice (Supplementary Figure S9A–D). NAD⁺/NADH and NADP⁺/NADPH ratios remained decreased in Cre⁺ female mice when compared to controls (Supplementary Figure S9E) On the other hand, OB-specific CypD GOF in males in the 12Mo cohort led to higher changes in bone metabolic profile than at 4Mo. Important changes in amino acids, glycolytic and TCA cycle intermediates cooperated to highlight Warburg effect as the top hit in the metabolite enrichment analysis (Figure 7A–E). Interestingly, glycine and proline metabolism, and therefore collagen synthesis, rely on the interconnection of urea cycle, glutamate metabolism, and ammonia recycling pathways, also shown in the metabolite enrichment analysis (Figure 7B). This reinforces our finding that OB-specific CypD GOF decreased organic component of the matrix and proline content in bones from Cre⁺ males at 12Mo. Finally, oxidized to reduced glutathione ratio (GSH/GSSG) is decreased in Cre⁺ male mice vs Cre⁻ controls at 12Mo (Figure 7F), indicating high levels of oxidative stress.

Figure 7. Metabolic changes in bone tissue from osteoblast-specific CypD gain-of-function male mice at 12Mo.

Small metabolites were extracted from tibial bone shafts of Col1^CreERt2/+;R26^caPpif/+ and R26^caPpif/+ control male mice at 12Mo and analyzed using LC-MS. A) Volcano plot showing all captured metabolites as a function of log(P value) over log(fold change) between Cre⁺ and Cre⁻ samples; B) Pathways significantly changed in the Metabolite Enrichment Analysis, false discovery rate (FDR) set at <0.1. Metabolites grouped into appropriate metabolic pathways: C) Glycolysis; D) TCA cycle; E) Amino acids; F) Bioenergetic and RedOx Ratios. Plots show the actual data points from six to seven independent mice per group. Calculated means and P value vs Cre⁻ control mice determined by an unpaired t-test. Two and a half fold change over Cre⁻ mice was set as significancy level for metabolite changes.

In general, OB-specific CypD GOF leads to mitochondrial dysfunction in vivo sufficient to decrease bone organic matrix anabolism, shifts bone energy metabolism to higher glycolysis levels, increases FA synthesis, and presents clear signs of premature aging and oxidative stress. All these observations are modulated in a sex-dependent manner.

DISCUSSION:

In physiological conditions, e.g. in undifferentiated BMSCs and some other stem and progenitor cells, high levels of CypD and MPTP activity are associated with a low electron flow into the ETC and active glycolytic pathway^(19–21). OxPhos activation, as observed during differentiation of BMSCs and other progenitors, requires MPTP closure to establish the Δψm which is the driver of Δp⁽²²⁾. This phenomenon allows the respiratory chain to be coupled to ATP production and reducing and anaplerotic sources, such as glutamine, to enter into the TCA cycle⁽⁴¹⁾. MPTP inactivation is achieved by downregulation of Ppif expression, as previously described in neuronal, cardiomyocyte, and OB differentiation.^(19–22). Therefore, cells dependent on mitochondrial OxPhos should downregulate Ppif expression and MPTP activity to properly commit to their function. Not surprisingly, we have shown that Ppif deletion confers higher OxPhos activity and osteogenic potential to BMSCs protecting against bone degeneration in aging^{(10, 22)}. On the other hand, CypD overexpression and MPTP opening are associated with pathological conditions such as mitochondrial dysfunction and swelling, inflammation, and aging. In fact, mitochondrial dysfunction and decreased OxPhos parameters are important hallmarks of aging^{(11–14, 35)}.

In this work we used a unique OB-specific mouse model, previously created in our group, of CypD GOF showing increased MPTP opening activity⁽²²⁾. We found that higher CypD/MPTP levels decreases OB function along with OxPhos parameters and increases glycolytic basal rate in vitro. During bone maintenance, OB function was found impaired following CypD GOF but with some sexual dimorphism. OB activity also showed to be differently affected in terms of mineral or organic matrix deposition as a function of sex. These features were further characterized when we analyzed bone morphological and biomechanical properties. Vertebral bones of CypD GOF mice presented a steep inter-cohort decay in both sexes whereas organic matrix content and yield load in long bones were affected only in males. Metabolomic analysis of bone tissue from middle aged CypD GOF mice confirmed the glycolytic shift shown in vitro and also revealed various changes in metabolic pathways and metabolite levels characteristic of aged bones^{(1, 2, 10, 42)}.

Several studies have shown that OB differentiation is accompanied by mitochondrial and OxPhos activation, characterizing well-functioning OBs as OxPhos-dependent cells^(4–8). Indeed, we previously reported that improving OxPhos function promotes the osteogenic program in BMSCs and improves OB function in bone maintenance and repair^{(6, 10, 22, 25, 43)}. Not surprisingly, higher OxPhos in BMSCs correlates with better spinal fusion outcomes in both human patients and in a mouse model⁽⁴⁴⁾. Conversely, CypD GOF decreases Δψm, and therefore mitochondrial and OxPhos function, affecting osteogenic cascade activation and decreasing OB function⁽²²⁾, also confirmed in the present study. Additionally, we confirmed that changes in mitochondrial function caused by caPpif overexpression led to decreased OxPhos parameters and mitochondrial ATP production rate in OBs. As regularly observed in cells with dysfunctional mitochondria, we found increased mtDNA content in CypD GOF cells which is likely a compensatory effect^{(10, 27)}. CypD GOF OBs also failed to promote mitochondrial fusion during differentiation as observed in Cre⁻ control OBs and previously described by us⁽⁵⁾. With the exception of mitochondrial relative elongation, all morphological parameters for mitochondrial fusion and network formation were found decreased in CypD GOF OBs during differentiation. Although the basal levels of mitochondrial ROS did not increase in CypD GOF OBs, they were significantly increased when cells were challenged with H₂O₂. This indicates that when CypD/MPTP is more active, OBs cannot properly handle oxidative stress. Of important note, the focus of this study is Col1-expressing OBs, therefore we are targeting late-stage mature OBs. Matrix synthesis is an energy-consuming process and highly dependent on OxPhos function. It is very likely that metabolic preferences reflect the stage of OB differentiation which is variable depending on the source and OB differentiation stage, as recently discussed in our recent review⁽³⁾. Reports showing increased glycolytic activity during OB differentiation^(45–47) might have captured early stages of OB differentiation represented by a high proliferative rate, which is indeed, glycolysis-dependent⁽⁴⁸⁾. Additionally, the source of OBs can affect metabolic preferences, as shown in studies using primary calvarial cells^{(46, 47, 49–51)} and calvaria-derived MC3T3 cell line^{(8, 46)}. The effect of cell isolation is especially relevant to the primary calvarial cells since digestion decreases the level of sugars, TCA cycle metabolites, and shifts the metabolic profile towards a higher glycolytic pathway⁽³⁾. Similarly, the metabolic preference of immortalized cell lines is characterized by a higher glycolytic pathway when compared to their representative primary cells⁽⁴⁸⁾. On the other hand, in vivo evidence demonstrated that calvaria OBs, embedding OBs, and osteocytes show increased OxPhos function over glycolytic activity⁽⁵²⁾. Despite some disparities, the vast majority of recent studies detected activation of OxPhos during osteogenic differentiation, characterizing OBs as actively respiring cells regardless of their source ^{(4–6, 8, 22, 25, 43, 50–56)}. Accordingly, we showed that a decrease in OxPhos function in OBs due to caCypD overexpression is sufficient to impair OB function which is not recovered by the compensatory glycolic shift.

Bone phenotype is characterized by a pronounced sexual dimorphism. Estrogen is a major player in such differences affecting both osteoclast and OB activity⁽⁵⁷⁾. Estrogen, and more specifically estradiol, also has important effects on mitochondrial function and dynamics, protecting against MPTP opening and facilitating mitochondrial ATP synthasome assembly^{(58, 59)}. Females also present with higher pyruvate dehydrogenase (PDH) activity favoring OxPhos usage in exchange of aerobic glycolysis, as described in mouse brain tissue⁽⁵⁹⁾. We previously reported that manipulations in Ppif expression and MPTP activity have different effects on bone tissue during fracture repair in males and females⁽²⁵⁾. MPTP shows strong sensitivity to estrogen and, thus, in vivo female mitochondria may be better protected against CypD/MPTP gain-of-function.

In this work, our histological analyses demonstrate that sexual dimorphism is associated with a temporal and spatial-dependent OB function in OB-specific CypD GOF mice. This association is confirmed by bone morphological and biomechanical analyses. When caCypD is expressed in OBs, vertebral bones show accelerated decay of morphological and biomechanical properties in both sexes over time, as depicted by bone fractional volume measurements and biomechanical testing; whereas long bones are affected mostly in males. This is explained by the fact that middle-aged (12 Mo) mice are more sensitive to bone loss than young (4 Mo) mice due to higher levels of oxidative stress and inflammation ^(60–63). Additionally, long bones have a lower remodeling rate than vertebral bones ⁽⁶⁴⁾, which explains the higher sensitivity of vertebrae to CypD/MPTP gain of function.

OB function partially regulates osteoclast recruitment and maturation. In the 12Mo cohort, we observed decreased osteoclast activity for TRAP staining in OB-specific CypD GOF mice only in the femur, however, it was not significant for females. Anyway, this long bone-specific decrease in osteoclast activity characterizes a spatial-dependent cross-talk regulation under CypD/MPTP activity control. Accordingly, vertebral bones are described to be the first bones affected by bone loss in aging⁽³¹⁾. We can assume that the increase in CypD/MTPT activity and decline in cell metabolism present in aging, favor bone catabolism in vertebral bones over long bones. Moreover, vertebral biomechanical properties actually increase in CypD GOF females at 4Mo, indicating distinct temporal effects in function of sex. We can suggest that sex-dependent regulatory processes in cell energy metabolism, at least partially controlled by Ppif expression and MPTP activity, have an antagonistic pleiotropic effect. Nonetheless, it is worth pointing out that despite the beneficial effects of CypD GOF at 4Mo in females, vertebral biomechanical properties decrease sharply over time (12Mo vs 4Mo) in Cre⁺ mice compared to Cre⁻ controls, regardless of sex. With regards to unmineralized organic matrix structure, our in vitro data demonstrated an important decrease in collagen production by CypD GOF OBs. Osteoporotic fractured bones have significantly higher CO₃-² to unmineralized organic matrix ratio when compared to controls, demonstrating the importance of organic matrix for bone strength⁽⁶⁵⁾. Raman spectrometry revealed that differences in unmineralized matrix are observed only in males at 12Mo, indicating that females are better protected against the effects of metabolic deregulations in long bones.

Corroborating with our in vitro results, metabolomic analysis of bone tissue revealed a glycolytic shift in mice with OB-specific CypD GOF, characterized by higher lactate production combined with decreased pyruvate levels. Also characteristic of glycolytic shift and deregulated glycolysis in Cre⁺ Females at 4Mo of age, are metabolite level changes suggestive of active glycogenolysis, increased utilization of the Pentose Phosphate Pathway along with decreased levels of glucose-1-P, glucose-6-P, and fructose-6-P and increased levels of glucogenic amino acid. Consistently, Warburg effect was a common hit in the metabolite set enrichment analysis along with decreased NAD⁺/NADH ratio in CypD GOF mice. In agreement with our findings, significantly decreased OxPhos function and NAD⁺ levels are described in aging due to a higher pore activity⁽¹³⁾, also showed by us in aged bones⁽²²⁾. Upon increasing MPTP opening during aging, NAD⁺ is hydrolyzed in the IMM, further contributing to NAD⁺ depletion and slowing TCA cycle operation⁽⁶⁶⁾.

Interestingly, we also found elevated levels of 2-hydroxyglutarate (2HG). The enantiomer L-2HG is found in hypoxia-induced cells as result of glutamine reductive carboxylation⁽⁶⁷⁾. Our findings suggest that CypD GOF and subsequently higher MPTP activity shifts canonical glutaminolysis to glutamine reductive carboxylation impacting glutamate metabolism and its interconnected pathways, such as proline and other pathways showed in the metabolite set enrichment analysis like urea cycle and ammonia recycling pathways. Glutamine is reported to have a crucial role in OB differentiation and function⁽⁷⁾. Glutamine is also an important source for anaplerosis and amino acid metabolism in general, and is the sole source of nitrogen in the TCA cycle⁽⁶⁸⁾. Proline metabolism is highly dependent on glutamate metabolism and takes place inside mitochondria. Making up to 1/4 of collagen structure, proline is the second most abundant amino acid⁽⁶⁹⁾. These observations may explain why we found decreased collagen deposition in vitro and proline content in bones from OB-specific CypD GOF male mice.

The metabolomic analysis also provides some insights into the differences in the effect of OB-specific CypD GOF in males and females. At 4Mo, we detected increased levels of kynurenate, itaconate, and 3-hydroxybutyrate, metabolites commonly associated with antiinflammatory effects, only in bones of OB-specific CypD GOF female mice. Kynurenate is a metabolite in the tryptophan pathway, an aromatic amino acid shown to promote bone anabolism⁽³⁸⁾. Itaconate is required for the activation of anti-inflammatory transcription factors⁽³⁹⁾ and 3-hydroxybutyrate acts as a CypD inhibitor, which also is shown to ameliorate inflammatory process⁽⁴⁰⁾. However, none of these metabolites were found significantly changed in males or 12Mo female cohort, matching the antagonistic pleiotropic effect of Ppif overexpression seen in vertebral biomechanical properties. Metabolomic analysis of bone tissue revealed that CypD GOF induces higher oxidative stress only in males, whereas it decreases NADPH regeneration only in females. As discussed above, estrogen can protect against the effects of metabolic deregulations in bone caused by higher CypD/MPTP activity, and therefore aging. Accordingly, ovariectomized mice experience a sharp decrease in bone mineral density and biomechanical properties, which can be targeted by hormone replacement therapy⁽⁷⁰⁾.

Overall, our results substantiate that mitochondrial dysfunction caused by caPpif expression and MPTP activity directly contributes to metabolic changes observed in aging leading to bone fragility. To date, there are sufficient studies confirming that the dysfunctional mitochondria observed in aging are important contributors to bone degeneration, and that improving mitochondrial stress-response and OxPhos function can mitigate the effects of aging on bone tissue. Herein, we demonstrate the causal relationship between higher CypD/MPTP activity and bone decay in aging, adding to our previous reports that CypD manipulation has a direct effect on bone phenotype. Moreover, we present evidence of a cell-autonomous response, directly connecting OB activity to mitochondrial energy metabolism and further demonstrating that proper OB function relies on mitochondrial and OxPhos-dependent mechanisms.

METHODS

Mouse strains:

CypD GOF mice, R26^caPpif with C57BL/6 genetic background, were created by the combined effort of our lab and Dr. George Porter’s lab in the University of Rochester Gene Targeting and Transgenic Core Facility, as previously described⁽²²⁾. In brief, the CTV vector reported by Dr. Changchun Xiao was used to insert caPpif encoding constitutively active K166Q CypD mutant. CaPpif cDNA containing Myc-DDT(Flag) tag on its N-terminal followed by IRES-eGFP was inserted at the Rosa26 locus. NeoSTOP cassette flanked by loxP sites was used to allow Cre inducible/temporal recombination. R26^caPpif mice were crossed to OB-specific tamoxifen-inducible Col1^CreERt2 mice, 2.3kb variant (final cross is Col1^CreERt2/+;R26^caPpif/+). 2.3kb Col1^CreERt2 mice (RRID:IMSR_JAX:016241), originally described by Kim JE, et al. 2004⁽⁷¹⁾, were a kind gift from Dr. Ackert-Bicknell (formerly of University of Rochester). Animal husbandry and experiments were performed upon approval of University of Rochester Institutional Animal Care and Use Committee (IACUC), and in accordance with state and federal law, and National Institutes of Health policy. All mice were housed at 23°C on a 12-h light/dark cycle with free access to water and PicoLab Rodent Diet 20 (LabDiet #5053, St. Louis, MO). Mice were in group housing when applicable based on weaning. Testing naïve mice with an average weight of 28 g were used for experiments. The assessments of animal studies were performed in a blinded and coded manner.

Isolation of BMSCs and osteoinduction:

Primary bone marrow cells were harvested from femurs and tibiae bone marrow from Col1^CreERt2/+;R26^caPpif/+ and Cre negative control littermates. Cells were plated at a density of 20×10⁶ per 10 cm dish in physiological ‘low’ glucose (5 mM) DMEM (LG-DMEM) supplemented with 1 mM L-glutamine, 10% fetal bovine serum (FBS), 1% penicillin-streptomycin, and incubated at 37 °C, 5% CO₂, and physiologically relevant 5% O₂. BMSCs were selected by plastic adherence, and further purified by at least two passages with short digest with 0.25% trypsin/1 mM EDTA between passages, as previously described⁽⁷²⁾. This procedure depletes cd45+ cells that require longer digest than BMSCs. Cells were osteoinduced at confluency in their appropriate media supplemented with 50 μg/mL ascorbate (TCI A2521) and 2.5 mM β-glycerolphosphate (USB Corp Cleveland, OH, 21655) for 11 and 14 days. OB differentiation was assessed by Alkaline Phosphatase (Thermo NBT/BCIP 1-step 34042 kit) and Alizarin Red staining as previously described⁽⁶⁾. In brief, staining intensity is measured in ImageJ, and mean value/ staining signal per well is annotated. Staining signal is then normalized to the cell number using the nuclear staining Hoechst33342 as described below. Our previous works showed strong correlation of staining intensity with gene expression^{(5, 22, 25)}.

In vitro collagen staining:

Sirius Red staining was performed as previously described⁽⁷³⁾. In brief, cells were first stained with the nuclear stain, Hoechst 33343 (Molecular Probes H 1339), at 0.5μM and a cell count/well performed by Celigo Image Cytometer using the fluorescent blue channel. Before the actual Sirius Red staining, cells were washed with cold 1X PBS three times. Next, 2 mL of Bouin’s solution was added per well and incubated for 1 hour at room temperature. Bouin’s solution was removed and cells were washed with running cold tap water on edges of wells to not disrupt the cells until yellow tint was gone. Plates were placed in fume hood to air-dry overnight. Two mL of 0.1% Sirius Red in saturated picric acid (Rowley Biochemical, SO-674) was added per well and incubated for 1 hour at room temperature. After incubation with Sirius Red, the staining solution was removed and 0.01M HCl was used to wash each well 4 times to remove excess dye. Plates were then photographed for visual reference. Two mL of 0.1M NaOH was used to elute the Sirius Red dye on a shaker for 30 minutes. The absorbance at 550nm while using 0.1M NaOH was measured as a control for a blank. The absorbance signal was normalized to cell number from Celigo count.

Real-time RT-qPCR:

Total RNA was isolated using the RNeasy kit (Qiagen 74106) and reverse transcribed into cDNA using the qScript cDNA synthesis kit (Quanta 95048–500). cDNA was subjected to real-time RT-PCR. The primer pairs used for genes of interest are outlined in Key Resources Table. Real-time RT-PCR was performed in the RotorGene system (Qiagen) using SYBR Green (Quanta 95072–012). Bglap and Col1a gene expression was normalized to B2m and used to confirm osteogenic differentiation.

Key Resources Table

Reagent type (species) or resource	Designation	Source or reference	Identifiers	Additional information
Genetic reagent (Mus musculus, male)	R26 caPpif/+	eLife11:e75023		Global CypD GOF or caPpif mouse (C57Bl/6 background)
Genetic reagent (Mus musculus, male)	2.3kb Col1CreERt2 (Tg(Col1a1-cre/ERT2)1Crm)	RRID:IMSR_JAX:016241		Col1-specific Cre inducible (C57Bl/6 background)
Genetic reagent (Mus musculus, male)	2.3kb Col1CreERt2;R26caPpif/+	eLife11:e75023		Col1-specific CypD GOF or caPpif mouse (C57Bl/6 background)
Transfected construct (Enterobacteria phage P1)	Ad5-CMV-Cre	Vector Biolabs	No:1045	Viral particle construct to transfect and express Cre recombinase
Biological sample (Mus musculus)	Primary BMSCs from R26caPpif/+ mice	This paper; eLife11:e75023		Freshly isolated from Mus musculus
Biological sample (Mus musculus)	Primary BMSCs from 2.3kb Col1CreERt2;R26caPpif/+ mice	This paper; eLife11:e75023		Freshly isolated from Mus musculus
Antibody	anti-CypD (mouse monoclonal) a.k.a Cyclophilin F	Abcam	RRID: AB_10864110	(1:1000)
Antibody	anti-VDAC1 (mouse monoclonal)	Santa Cruz Biotecnology	RRID: AB_632587	(1:2000)
Antibody	Goat anti-mouse (goat polyclonal, HRP conjugate)	Abcam	RRID: AB_478283	(1:3000) to (1:5000)
Antibody	anti-osteocalcin (rabbit polyclonal)	Enzo Life Sciences	RRID: AB_10540992	(1:400)
Antibody	anti-GFP (rabbit polyclonal)	Abcam	RRID: AB_303395	(1:500)
Antibody	goat anti-rabbit IgG (goat polyclonal, Alexa Fluor®647 conjugate)	Abcam	RRID: AB_2722623	(1:2000)
Antibody	goat anti-rabbit IgG (goat polyclonal, Alexa Fluor®488 conjugate)	Abcam	RRID: AB_2630356	(1:2000)
Recombinant DNA reagent	CAG-STOP-eGFP-ROSA26TV (plasmid)	Addgene	# 15912	Plasmid construct for gene knock-in engineering
Recombinant DNA reagent	caPpif	Bochaton et al, 2015		caCypD construct (CypD GOF)
Sequence-based reagent	B2m_Fwd	IDT	RT q-PCR primer	AATGGGAAGCCGAACATAC
Sequence-based reagent	B2m_Rev	IDT	RT q-PCR primer	CCATACTGGCATGCTTAACT
Sequence-based reagent	Col1_Fwd	IDT	RT q-PCR primer	CCAAACTCAGAAGATGTAGGAG
Sequence-based reagent	Col1_Rev	IDT	RT q-PCR primer	CATCATAGCCATAGGACATCT
Sequence-based reagent	Bglap_Fwd	IDT	RT q-PCR primer	GACCTCACAGATGCCAAG
Sequence-based reagent	Bglap_Rev	IDT	RT q-PCR primer	CAAGCCATACTGGTCTGATAG
Sequence-based reagent	18s_Fwd	IDT	RT q-PCR primer	TAGAGGGACAAG TGGCGTTC
Sequence-based reagent	18s_Rev	IDT	RT q-PCR primer	CGCTGAGCCAGTCAGTGT
Sequence-based reagent	Cox3_Fwd	IDT	RT q-PCR primer	CGAAACCACATAAATCAAGCCC
Sequence-based reagent	Cox3_Rev	IDT	RT q-PCR primer	CTCTCTTCTGGGTTTATTCAGA
Sequence-based reagent	Cre_Fwd	IDT	PCR primer	CCTGGAAAATGCTTCTGTCCGTTTGCC
Sequence-based reagent	Cre_Rev	IDT	PCR primer	GAGTTGATAGCTGGCTGGTGGCAGAT
Sequence-based reagent	caCypD (TetIRES)_Fwd	IDT	PCR primer	AATGGCTCTCCTCAAGCG
Sequence-based reagent	caCypD (TetGFP)_Rev	IDT	PCR primer	GCGGATCTTGAAGTTCACCTTGATGCCGT
Commercial assay or kit	Wizard SV DNA purification kit	Promega	A2360
Commercial assay or kit	Rat / Mouse PINP ELISA Kit	Immune Diagnostic Systems	AC-33F1
Commercial assay or kit	RatLaps CTX-I ELISA Kit	Immune Diagnostic Systems	AC-06F1
Chemical compound, drug	L-Ascorbic Acid 2-Phosphate Sesquimagnesium Salt	TCI America	TCI A2521
Chemical compound, drug	Alkaline Phosphatase Substrate	Thermo Fischer	34042	1-Step™ NBT/BCIP Substrate Solution
Chemical compound, drug	Sirius Red	Rowley Biochemical	SO-674
Chemical compound, drug	Hoechst 33343	Molecular Probes	H 1339
Chemical compound, drug	MitoTracker Red CMXRos	Invitrogen	M7512
Chemical compound, drug	MitoSox Red	Invitrogen	M36008
Chemical compound, drug	Tamoxifen	Sigma-Aldrich	T5648
Chemical compound, drug	4-OH-tamoxifen	Sigma-Aldrich	H7904
Chemical compound, drug	Hydrogen Peroxide	Sigma-Aldrich	H1009
Chemical compound, drug	IRDye 608 Bone Tag	LI-COR	926–09374
Chemical compound, drug	4% paraformaldehyde solution	Thermo Fischer	J19943-K2
Chemical compound, drug	Calcein	Sigma-Aldrich	C0875–5G
Chemical compound, drug	Oligomycin	Sigma-Aldrich	O4876
Chemical compound, drug	Antimycin A	Sigma-Aldrich	A8674
Chemical compound, drug	Rotenone	Sigma-Aldrich	R8875
Chemical compound, drug	2-deoxyglucose	Sigma-Aldrich	D3179
Chemical compound, drug	FCCP	Abcam	Ab120081

Seahorse metabolic profiling assay:

oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were measured using Seahorse XF96 (Seahorse Bioscience). BMSCs from Col1^CreERt2/+;R26^caPpif/+ mice were plated on Seahorse 96-well plates 48 hours before the experiment at a density of 20,000 cells per well. Cells were incubated in osteogenic media to induce osteogenesis for 14 days. Tamoxifen (4OH-Tam) was added on Day 9 of osteoinduction to trigger recombination and transgene expression in Cre⁺ cells. Cre⁻ cells were also treated with 4OH-Tam to account for possible effect of tamoxifen itself. The timing was similar to that used in our previous work and was based on the fact that Col1 gene and, therefore Col1^Cre becomes active around Day 9 in the in vitro system⁽²²⁾. Immediately before the experiment, media was replaced with unbuffered DMEM (Gibco A1443001) media containing 1 mM L-glutamine (Gibco 25030–081), 5 mM D-glucose (Sigma G8270) and no pyruvate (pH 7.4). A baseline measurement of OCR and ECAR was taken, and then an inhibitory analysis was performed using serial injections of 1 μg/mL oligomycin (Oligo), 1 μM FCCP, and simultaneous addition of 2 μM each of antimycin A (AntA) and rotenone (ROT). A parallel experiment was run with the following injections of Oligo + FCCP followed by 10 mM 2-deoxyglucose (Sigma D-3179). After analysis, total cell count was measured with 2 μg/ml Hoechst 33342 (Invitrogen H1399) using the Celigo image cytometer. The following OxPhos and glycolytic indexes were calculated: basal respiration (OCR_pre-Oligo - OCR_{post-AntA/ROT}), ATP-linked respiration (OCR_pre-oligo - OCR_post-Oligo), mitochondrial ATP production rate ((OCR_pre-Oligo - OCR_post-Oligo) × 2 (pmol O/pmol O₂) × 2.75 (P/O ratio, pmol ATP/pmol O)), maximal respiration (OCR_post-FCCP - OCR_{post-AntA/ROT}), proton leak (OCR_post-Oligo - OCR_{post-AntA/ROT}) and basal glycolysis (ECAR_pre-Oligo).

Mitochondrial DNA assay:

cells were collected and total DNA isolated using the Wizard SV DNA purification kit (Promega). Quantitative real-time PCR was used to detect the mtDNA-encoded mt-Cox3 (Cytochrome c oxidase subunit III gene) and nuclear-encoded 18s. The primer pairs used for genes of interest are outlined in Key Resources Table. Real-time PCR was performed in the RotorGene system (Qiagen) using SYBR Green (Quanta 95072–012). MtDNA was normalized to gDNA.

Mitochondrial functional assay:

7×10⁴ BMSCs from R26^caPpif/+ (CypD GOF) mice were plated on laminin-coated 25 mm round glass coverslips and infected with Ad5-CMV-Cre or Ad5-CMV-eGFP control virus at 500 MOI. Forty-eight hrs after infection, BMSCs were induced in osteogenic media. At day 0 or day 11 of osteoinduction, cells were stained with the nuclear stain, Hoechst 33343 (Molecular Probes H 1339), at 0.5μM and MitoTracker Red CMXRos (Invitrogen M7512) at 50nM or the mitochondrial ROS-dependent MitoSox Red (Invitrogen M36008) at 2 μM. MitoSox-stained cells were challenged with 1mM H₂O₂ (Sigma-Aldrich H1009). Cells were imaged using inverted fluorescence microscope (Axioscope) and fluorescence signal was quantified using ImageJ. For mitochondrial network analysis, CMXRos-stained cells had mitochondria identified in ImageJ using a convolution filter and thresholding for intensity and minimum pixel area as previously described⁽⁷⁴⁾. Particle analysis detects and calculates measurements for individual mitochondrion.

Western Blot:

Cell lysates and gel electrophoresis followed by transfer to polyvinylidene difluoride membrane were performed as previously described⁽²²⁾. Antibodies were diluted in 2.5% dry milk in PBST. For CypD detection, blots were probed with monoclonal CypD antibody (RRID: AB_10864110) diluted 1:1000 and HRP-conjugated goat anti-mouse antibody diluted 1:3000. To verify equal loading and considering that CypD is a mitochondrial protein, blots were re-probed with the commonly used for this purpose mitochondrial VDAC1 protein (RRID: AB_632587) antibody diluted 1:2000 and HRP-conjugated goat anti-mouse antibody diluted 1:5000. Antibody signal was developed with West Pico Substrate and imaged in BioRad imager.

Serum Biomarkers:

CTX-I and P1NP ELISA kits (Immune Diagnostic Systems) were used to measure osteoclast and osteoblast activity, respectively. Blood was collected from submandibular vein and serum prepared. Samples were run according to the manufacturer’s instructions.

In Vivo Imaging System (IVIS^®):

mice hip area and lower limbs were shaved, then mice were injected with 0.1ml (equivalent to 0.08 nmol/g for a 25g mouse) reconstituted IRDye 608 BoneTag (LI-COR) IP, 48 hrs prior to imaging. Under isoflurane anesthesia, mice are imaged and infra-red fluorescence measured using IVIS Spectrum (Caliper Life Sciences). A negative control mouse, which received only PBS IP injection was used to subtract non-specific infra-red signal arising from bone autofluorescence.

Dual-Energy X-ray Absorptiometry (DEXA):

mice were anesthetized using 100 mg/kg Ketamine and 10 mg/kg Xylazine IP at a rate of 0.1ml per 10 grams of body weight. Under deep anesthesia, mouse whole body is scanned and area of interest selected using Lunar PIXImus2 system. For our purpose, whole body and L6–L5 are analyzed in terms of bone mineral density (BMD), bone mineral content (BMC) and fat percentage.

Histology:

After micro-CT analysis, bone specimens were NBF-fixed and processed for histology via decalcification in Webb-Jee 14% EDTA solution for one week followed by paraffin embedding. Sections were cut to 5 μm in three levels of each sample, and then stained with either TRAP or immunofluorescence (IF).

Immunofluorescence:

NBF-fixed bone specimens were processed as above. Immunofluorescence (IF) was carried out using a primary anti-osteocalcin antibody (RRID:AB_10540992) diluted 1:400 or anti-GFP antibody (RRID:AB_303395) diluted 1:500, followed by incubation with anti-rabbit IgG secondary antibody conjugated with Alexa Fluor^®647 (RRID:AB_2722623) diluted 1:2000 or with anti-rabbit IgG secondary antibody conjugated with Alexa Fluor^®488 (RRID:AB_2630356), respectively. The primary antibody solution was composed of PBS, 0.1% Tween-20 and 5% Goat Serum. Prior incubation at 65°C in 10mM sodium citrate (pH 6.0) for 3 hrs was performed for antigen retrieval. Fluoroshield Mounting Medium with DAPI (ab104139) was used to counterstain and coverslip IF-slides.

Bone dynamic histomorphometry:

as previously described⁽¹⁰⁾, mice were labeled via intraperitoneal injection with 25mg/Kg of 0.4% calcein (Sigma, # C0875–5G) solution at day 0 followed by a second calcein injection at day 10. Seven days after the second calcein injection, mice were sacrificed. Tibiae were collected, grossed, fixed in 4% paraformaldehyde solution (Thermo Scientific, J19943-K2), and processed for frozen sectioning. Sections were visualized using a fluorescence Axioscope 40 microscope (Zeiss) equipped with an Olympus DP72 camera (Olympus) and analyzed using ImageJ software to calculate bone formation rate (BFR) and mineral apposition rate (MAR) as follows: BFR = MAR × (MS/BS) where MAR = Ir.L.Th/Ir.L.t. and MS = (dLS+sLS/2)/BS (BS, bone surface; dLS, double labelled surface; sLS, single labelled surface; Ir.L.Th, distance between labels; and Ir.L.t, time between labels). Three different levels were counted per mouse and averaged.

Histomorphometry:

TRAP or IF-stained slides were scanned in an Olympus VSL20 whole slide imager at 40x magnification and evaluated with VisioPharm automated histomorphometry software. TRAP-stained slides were analyzed to measure the TRAP-positive area relative to the total bone area in the bone specimens. IF-stained slides probed for osteocalcin were analyzed to measure the fluorescence signal intensity relative to the total bone area. Both bone lining and embedded OCN⁺ cells were included in the quantification procedure taking into account their proportional relative number within age groups. EGFP IF-stained slides had GFP signal quantified as described above but further normalized to cell count to equalize the differences in cell number per bone surface among different age groups. Lining and embedded eGFP+ cells were equally included per slide. Since no differences in eGFP expression were found between long and vertebral samples, the results combined both bone types (n=3 vertebrae + n=3 tibiae) from different mice. Two to three different levels were counted per sample/specimen and averaged.

Bone micro-computed tomography (μCT):

following euthanasia, femur and vertebrae were isolated and cleaned of excess soft tissue. Bones were stored at −80 °C prior to μCT. Specimens were imaged using high resolution acquisition (10.5 μm voxel size) with the VivaCT 40 tomograph (Scanco Medical). Scanco analysis software was utilized for volume quantification.

Biomechanical testing:

Compression test:

vertebrae samples (L4 and L3) are held in bone cement and compressed using EnduraTec TestBench system (Bose) at a displacement rate of 1mm per minute until crushing. Force data are plotted vs deformation to determine the maximum force and energy to maximum. Three-point bending test: immediately following μCT scanning, specimens were subjected to biomechanical testing. Femur is placed horizontally upon two points and the stress applied to the top of the sample through a single point at 0.03mm/second displacement rate until failure. Young’s modulus, maximum stress and energy to maximum are calculated through plotted stress/strain analysis plus moment of inertia and maximum radius perpendicular to the neutral plane from prior μCT analysis.

Raman spectroscopy:

as described previously^{(34, 75)}, Raman measurements were performed ex vivo on the excised, thawed humeri using an 830 nm, 150 mW laser that was focused in a spot with 0.5-mm diameter for 300 seconds. Five locations were measured and averaged along the anterior mid-diaphysis in 1 mm intervals. To control for collagen fiber orientation, the bone orientation was kept consistent for all measurements. Excitation and detection were performed in a non-contact manner. Spectral pre-processing involved cosmic ray removal, readout and dark current subtraction, and image aberration correction. The spectra (744–1740 cm⁻¹) were fit to a 5th order polynomial to remove fluorescence. Spectra were smoothed with a Savitzky-Golay filter⁽⁷⁶⁾ over a window of 3 pixels (5 cm⁻¹). Spectra were normalized to their mean absolute deviation (MAD)⁽³⁴⁾. Commonly reported ratios were calculated by taking the ratio of the area under the curves for specific spectral peaks. Such ratios included: phosphate mineral to matrix ratio, defined by the phosphate (924–986 cm⁻¹) and matrix defined by either proline (836–868 cm⁻¹), amide III (1218–1362 cm⁻¹), CH₂ (1410–1496 cm⁻¹), or amide 1 (1596–1730 cm⁻¹), carbonate mineral (1054–1098 cm⁻¹) to matrix, and carbonate to phosphate. These peak areas are modified from those used in Unal et al.⁽⁷⁷⁾ and Hammond et al.⁽⁷⁸⁾ to match our Raman spectrum. The full width half maximal (FWHM) bandwidth of the phosphate peak, which is related to the mineral crystallinity ^{(79, 80)} was calculated by fitting the peak with a single Gaussian curve. Principal component analysis (PCA) was used to reduce the dimensions of the data. Specifically, our data consisted of 1 spectrum/mouse and the data was decomposed to principal components (PCs) using MATLAB.

Metabolomics:

samples are prepared as previously described⁽⁵⁾. Tibial bone shafts were cleaned of bone marrow, soft tissue, and cartilage; frozen in liquid nitrogen, and pulverized. Metabolites were extracted in 4 ml of 80% methanol, dried under nitrogen stream and reconstituted in 20ul of 50% methanol. Samples were analyzed using reverse phase liquid chromatography (LC) with an ion-pairing reagent in a Shimadzu HPLC coupled to a Thermo Quantum triple-quad mass spectrometer (MS). Significancy level was set as fold change below or above 2.5 and P value below 0.5. Metabolite enrichment analysis was done using MetaboAnalyst web server and false discovery rate set below 0.1^{(81, 82)}.

Statistics:

A power analysis on normalized biomechanical data was performed since it showed the highest variance. It was determined that some biomechanical and metabolomic quantitative outcomes would require 7 mice per group, and quantitative histological results would require 5 mice per group. We set the significance level at 5% (α=0.05) and Type II error (β) to ≤20%. For statistical analysis, we compared the difference between two simple groups independently, therefore an unpaired t-test was used when the frequency distribution of the differences between the two groups fitted a normal distribution. Although we analyzed independent variables and therefore independent hypothesis in the multi-group graphs, we performed Ordinary one-way ANOVA using Dunnett’s multiple comparisons test with a single pooled variance to further validate our statistical findings significance. Since no differences in significance were found, we maintained our t-test results. For actual multicomparison testing, we utilized two-way ANOVA when changes among sex groups (male vs female) and age cohorts (4Mo vs 12Mo) were simultaneously analyzed.

Supplementary Material

supinfo

Supplementary Figure S1. CypD gain-of-function impairs mitochondrial ATP rate and network in osteoblasts. BMSCs from Col1^CreERt2;R26^caPpif/+ or R26^caPpif/+ control mice were cultured in osteogenic media for 11 to 14 days. A) BMSCs from Col1^CreERt2;R26^caPpif/+ mice osteoinduced for 14 days presented decreased mitochondrial ATP production rate when compared to Cre⁻ control cells. B) Mitochondrial DNA expression normalized to genomic DNA is increased in Cre⁺ cells; Osteoinduced R26^caPpif/+ BMSCs infected with Adeno-Cre, cultured in osteogenic media for 11 days and stained with CMXRos showed decreased mitochondrial morphology and network parameters for Cre⁺ cells. Mitochondria are identified in imageJ using a convolution filter and thresholding for intensity and minimum pixel area. Particle analysis detects and calculates measurements for individual mitochondrion; C) Mitochondrial area; D) Mitochondrial perimeter; E) Mitochondrial length; F) Mitochondrial width; G) Mean form factor; H) Mean aspect ratio. Plot shows the actual data points (biological replicates), calculated means and P value vs Cre⁻ controls determined by an unpaired t-test. For Seahorse assay, n=15 (three biologic replicates of five technical replicates). For mitochondrial network analysis, n=9, a total of 6578 mitochondria were analyzed in the Cre⁻ group and 2499 mitochondria were analyzed in the Cre⁺ group.

Supplementary Figure S2. BMSCs from CypD gain-of-function mice show decreased mitochondrial networking. BMSCs were collected from R26^caPpif/+ mice and recombination induced through Adeno-Cre viral infection. Cells were cultured in regular media, stained with the nuclear stain, Hoechst33342, and MitoTracker red CMXRos. A) Representative image of undifferentiated Adeno-Cre-infected R26^caPpif/+ BMSCs showing both transduced (GFP⁺) and non-transduced control (GFP⁻) cells. Mitochondria are identified with ImageJ using a convolution filter and thresholding for intensity and minimum pixel area. Particle analysis detects and calculates measurements for individual mitochondrion; B) Number of mitochondrial endpoints per cell normalized to total mitochondrial area; C) Mitochondrial area; D) Mitochondrial perimeter; E) Mitochondrial length; F) Mitochondrial width; G) Mean form factor; H) Mean aspect ratio. Plot shows the actual data points (biological replicates). Calculated means and P value vs Cre⁻ controls determined by an unpaired t-test. A total of 14516 mitochondria were analyzed in the Cre⁻ group and 14844 mitochondria were analyzed in the Cre⁺ group.

Supplementary Figure S3. Successful recombination in the bones of OB-specific CypD gain-of-function mice. Recombination and therefore, CypD GOF was induced with Tamoxifen in 2Mo-old Col1^CreERt2/+;R26^caPpif/+ mice. Tissue was harvested at 4Mo or 12Mo preceded by bimonthly boost injections. A) Representative image of immunofluorescence eGFP signal confirming the successful recombination in the tibiae of Cre⁺ mice in the 4Mo and 12Mo cohorts. Representative images from tibiae metaphysis (BM: bone marrow). No differences in eGFP signal were found between groups and timepoints; B) Bone marrow stromal cells (BMSCs) from either osteoblast (OB)-specific, inducible 2.3 kb Col1^CreERt2;R26^caPpif/+ mice or R26^caPpif/+ mice were cultured in osteogenic media for 7–14 days. Representative western blot of BMSCs from 2.3 kb Col1^CreERt2;R26^caPpif/+ mice, no caCypD is detected in the earlier stage (Day 7) of OB differentiation. C) and D) Representative images of BoneTag infrared signal captured by IVIS in the tail proximal region or tibiae. Signal quantification showed no differences in OB activity for both vertebrae (C) and tibiae (D) at 4Mo; E) representative sections and quantitative analyses of mouse bones labeled with Calcein for dynamic Mineral Apposition Rate (MAR), and Bone Formation Rate (BFR) assay. Dynamic histomorphometry analysis revealed no differences in OB activity when Cre⁺ and Cre⁻ control mice were compared in the 4Mo cohort. F) Serum biomarker CTX-1 for OC activity showed no differences between Cre⁺ and Cre⁻ mice. Representative images of TRAP-stained sections from vertebrae (G), and tibiae (H). TRAP staining quantification revealed no differences between Cre⁺ and Cre⁻ mice. Plots show the actual data points from four to ten independent mice per group. Males and females were analyzed separately to account for sexual dimorphism. Calculated means and P value vs Cre⁻ control mice determined by an unpaired t-test or Two way-ANOVA for multicomparison test.

Supplementary Figure S4. CypD gain-of-function decreases osteoblast activity in vivo. Recombination and therefore, CypD GOF was induced with Tamoxifen in 2Mo-old Col1^CreERt2/+;R26^caPpif/+ mice. Tissue was harvested at 12Mo preceded by bimonthly boost injections. A) Serum biomarker for OB activity, P1NP, showed no differences between experimental and control groups at 12Mo; B) and C) Representative images of BoneTag infrared signal captured by IVIS in the tail proximal region or tibiae. Signal quantification showed no differences in OB activity in the tibiae (C), but decreased signal in the vertebrae (B) of Cre⁺ mice at 12Mo when compared to Cre⁻ control mice; D) Serum biomarker CTX-1 for OC activity showed no differences between Cre⁺ and Cre⁻ mice. Representative images of TRAP-stained sections from vertebrae (E), and tibiae (F). TRAP quantification revealed no differences in OC activity in the vertebrae in Cre⁺ vs Cre⁻ mice but decreased TRAP staining in the tibiae of Cre⁺ mice at 12Mo. Plots show the actual data points from seven to twelve independent mice per group. Males and females were analyzed separately to account for potential sexual dimorphism. Calculated means and P value vs Cre⁻ control mice were determined by an unpaired t-test.

Supplementary Figure S5. Bone density and volumetric parameters in mice with osteoblast-specific CypD gain-of-function in the 4Mo and 12Mo cohorts. Bone density and volumetric properties of vertebrae from Col1^CreERt2/+;R26^caPpif/+ and control R26^caPpif/+ mice were measured by DEXA and μCT respectively. No differences in the following bone morphological parameters were found between Cre⁺ and Cre⁻ control groups in 4 and 12Mo cohorts: A) Body mass; B) Bone mineral density (BMD) from whole skeleton; C) BMD from vertebrae L6 and L5. D) and E) higher fat accumulation is present within the whole skeleton and lumbar region (vertebrae L6+L5) in females in the Cre⁺ group when compared with Cre⁻ control group at 4Mo. The 12 Mo cohort showed no differences in fat accumulation between experimental and control groups in both males and females. No differences in the following μCT measurements were found between Cre⁺ and Cre⁻ groups in the 4Mo cohort: F) Cortical BMD, G) Trabecular BMD, and G) Trabecular space. However, 12Mo cohort showed increased Tb.BMD for females whereas it was decreased for males. Plots show the actual data points from eight to twelve independent mice per group. Males and females were analyzed separately to account for potential sexual dimorphism. Calculated means and P value determined by an unpaired t-test.

Supplementary Figure S6. Bone density, biomechanical properties, and organic matrix content in long bones in mice with osteoblast-specific CypD gain-of-function. Bone density, biomechanical properties, and chemical composition of Col1^CreERt2/+;R26^caPpif/+ and R26^caPpif/+ control mice were measured by μCT, a three-point bending test, and Raman spectrometry, respectively. A) Cortical BMD in Cre⁺ vs Cre⁻ mice is decreased in the 4Mo cohort but increased in the 12Mo cohort; B) Trabecular BMD is decreased only in males at 4Mo when comparing Cre⁺ vs Cre⁻ mice; No differences on the following bone biomechanical parameters were found between Cre⁺ and Cre⁻ control groups in 4 and 12Mo cohorts: C) Young’s modulus and; D) Maximum load. Chemical composition of the humerus from Col1^CreERt2/+;R26^caPpif/+ and control R26^caPpif/+ mice at 12Mo was detected with Raman spectrometry: E) Carbonate (CO₃⁻²)/Phosphate (PO₄⁻³) ratio, PO₄⁻³/amide III (F), and CO₃⁻²/amide III (K) ratios show no differences between experimental and control groups; Plots show the actual data points from eight to twelve independent mice per group. Males and females were analyzed separately to account for sexual dimorphism. Calculated means and P value vs Cre⁻ control mice determined by an unpaired t-test.

Supplementary Figure S7. No differences in mineral and organic matrix content were found in long bones from osteoblast-specific CypD gain-of-function mice at 4Mo. Average spectra regarding the chemical composition of humerus from Col1^CreERt2/+;R26^caPpif/+ and R26^caPpif/+ control mice at 4Mo were measured by Raman spectrometry. A) Table depicting Raman measurements for males; B) Table depicting Raman measurements for females. Table show the measurement mean and standard deviation (SD) from seven independent mice per group. Males and females were analyzed separately to account for sexual dimorphism. Calculated means and P value vs Cre⁻ control mice determined by an unpaired t-test.

Supplementary Figure S8. Metabolic changes in bone tissue from osteoblast-specific CypD gain-of-function mice at 4Mo. Small metabolites were extracted from tibial bone shafts from Col1^CreERt2/+;R26^caPpif/+ and R26^caPpif/+ control female mice and analyzed using LC-MS. A) and D) Pathways significantly changed in the Metabolite Enrichment Analysis. False discovery rate (FDR) set at <0.1; Metabolites grouped into appropriate metabolic pathways with more significant changes observed in females: B) and E) Amino acids; C) and H) Anti-Inflammatory; F) Gluconeogenesis; G) Fatty acid synthesis. Plots show the actual data points from six to seven independent mice per group. Males and females were analyzed separately to account for sexual dimorphism. Calculated means and P value vs Cre⁻ control mice determined by an unpaired t-test. Two and a half fold change over Cre⁻ mice was set as significancy level for metabolite changes.

Supplementary Figure S9. Moderate metabolic changes are detected in bone tissue from osteoblast-specific CypD gain-of-function female mice at 12Mo. Small metabolites were extracted from tibial bone shafts from Col1^CreERt2/+;R26^caPpif/+ and R26^caPpif/+ control female mice and analyzed using LC-MS. A) Volcano plot showing all captured metabolites as a function of log(P value) over log(fold change) between Cre⁺ and Cre⁻ samples. Metabolites grouped into appropriate metabolic pathways with few significant changes observed in: B) Glycolysis; C) TCA cycle; and D) Amino acids; E) Bioenergetic and RedOx Ratios, energy charge ([ATP + 0.5 ADP] /[ ATP + ADP + AMP]), reduced to oxidized glutathione ratio (GSH/GSSG), oxidized to reduced NADP ratio (NADP⁺/NADPH), and oxidized to reduced NAD ration (NAD⁺/NADH). Plots show the actual data points from six independent mice per group. Calculated means and P value vs Cre⁻ control mice determined by an unpaired t-test. Two and a half fold change over Cre⁻ mice was set as significancy level for metabolite changes.

The abbreviations used are:

2HG

2-hydroxyglutarate

AA

antimycin A

ALP

alkaline phosphatase

Ar

area

AR

alizarin red

ATP

adenosine triphosphate

BFR

bone formation rate

BM

bone marrow

BMD

bone mineral density

BMSC

bone marrow stromal (a.k.a. mesenchymal stem) cell

BV/TV

bone over total volume

caPpif

constitutively active Ppif

cKO

conditional knockout

Ct

cortical

CT

computed tomography

CypD

cyclophilin D

DEXA

dual-energy X-ray absorptiometry

ECAR

extracellular acidification rate

ETC

electron transport chain

FA

fatty acid

FCCP

carbonyl cyanide p-trifluoromethoxyphenylhydrazone

f/f

floxed/floxed

GOF

gain-of-function

IR

infra-red

IMM

inner mitochondrial membrane

IVIS

in vivo image system

K166Q

substitution of lysine at position 166 for glutamine

KO

knock-out

LC

liquid chromatography

LOF

loss-of-function

MAR

mineral apposition rate

MPTP

mitochondrial permeability transition pore

MS

mass spectrometry

N

number

NF

not found

OB

osteoblast

OC

osteoclast

Ocn

osteocalcin

OCR

oxygen consumption rate

Oligo

oligomycin

OT

osteocyte

OxPhos

oxidation phosphorylation respiration

PCA

principal component analysis

PDH

pyruvate dehydrogenase

PPP

pentose phosphate pathway

ROS

reactive oxygen species

ROT

rotenone

SBE

Smad binding element

Sp

space

SR

sirius red

Tb

trabecular

TCA

tricarboxylic acid

Th

thickness

TRAP

tartrate-resistant acid phosphatase

Δψm

mitochondrial membrane potential

Δp

proton motive force

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon request.