Bone marrow adipose tissue (BMAT), found mostly in the medullary canal of long bones, constitutes roughly 10% of total adipose tissue mass in humans (1) and comprises of adipocytes (BMAds) that are morphologically similar to white adipocytes (1, 2). However, studies suggest that BMAT is an adipose depot that is distinct from extramedullary adipose tissue (1–4). BMAT responds to changes in energy metabolism by modifying the size and amount of BMAds. In tibia, the basal characteristic of BMAds revealed two anatomically-distinct subpopulations in mice: loosely-packed small BMAds in the proximal tibia and a dense network of large BMAds in the distal tibia referred to herein as regulated and constitutive BMAds, respectively (5–7). Physiologically, constitutive BMAds are less responsive than regulated BMAds to external stimuli such as β-adrenergic stimulation (4) and caloric restriction (8), and are associated with cortical bone loss (9). Regardless, the general expansion of BMAds during both caloric surplus and deficit have been shown to affect local bone metabolism (10–12) and also global energy metabolism (1, 13). Metabolically, adult bone marrow (BM), comprising mostly of BMAT, has been shown to have high basal glucose uptake in humans, exceeding that of white adipose tissue (3, 13). This suggests that BMAT may play a role in whole-body glucose metabolism.

We have previously shown that BMAT-enriched femoral BM insulin-stimulated glucose uptake (GU) and fasting free fatty acid uptake are impaired in insulin resistance in humans (14). Further, lumbar vertebral BM metabolism is affected by body weight (15). We also showed that BMGU is affected by exercise training and the effects of short- and long-term exercise training on BMGU varied according to anatomical location (14, 15). Femoral insulin-stimulated BMGU was increased already after two weeks of cycling exercise and associated positively with whole-body insulin sensitivity and bone turnover rate at baseline. On the contrary, lumbar vertebral BMGU remained unaffected by short-term exercise training, but was higher in participants with high body weight while six months of regular exercise training decreased lumbar vertebral BMGU to the level of leaner participants even without reduction in body weight or BM adiposity (15).

In our previous studies in humans, we used positron emission tomography (PET) to assess the changes in BM metabolism (3, 14, 15). This imaging modality allowed us to study the metabolism of the BM niche as a whole. However, we could not differentiate between the different tissues of BM, including BMAT, trabecular bone and hematopoietic cells, or study the mechanisms underlying these changes. In this study, to gain insight into the effects and mechanisms of weight gain, insulin resistance, and exercise training on BM in more detail, we used PET/CT to evaluate humeral and sternal BMGU and histological analyses to assess tibial BM adiposity. Rats were fed either chow or high-fat diet (HFD) for 24 weeks. Additionally, after 12 weeks, HFD-fed rats switched either to chow diet, voluntary running exercise, or both for another 12 weeks. We hypothesize that HFD impairs insulin-stimulated BMGU and increases BM adiposity, and that these findings can be ameliorated by switch to chow diet and/or running exercise.

In this study, we examined BMGU using [¹⁸F]FDG-PET/CT during insulin stimulation and BM adiposity using immunohistochemistry in male Sprague-Dawley rats. Eight-week-old rats were exposed to either chow or HFD for 24 weeks. Additionally after 12 weeks, HFD-fed rats switched either to chow diet, voluntary running exercise, or both for another 12 weeks. A long-term 24-week exposure to HFD markedly increased body weight, whole-body adiposity and induced hypertrophy of regulated BMAds in the proximal tibia, consistent with previous findings in mice (11). The group with double intervention (¹²HFD+¹²(Chow+E)) had the most pronounced reduction in whole-body fat percentage when compared to ²⁴HFD and ²⁴Chow groups. HOMA-IR index, a measure of insulin resistance, was significantly lower in both groups that switched to chow diet compared to ²⁴HFD group which suggests an improved insulin sensitivity. Interestingly, voluntary wheel running during HFD did not affect the whole-body fat percentage when compared to the sedentary ²⁴HFD. HFD affected mainly the trabecular region of bone implying trabecular bone loss, which is likely due to higher bone metabolism in that region responding to the intervention, similar to what has been reported elsewhere (6, 24).

BMGU was higher in 8-week-old rats compared to 32-week-old rats in both humeral and sternal BM. This phenomenon may be explained by the function of BM and the physiological “red-to-yellow” transition. The BM cavity of young rodents as well as humans consists predominantly of hematopoietic tissue, which during aging is replaced by adipocyte-rich yellow BM (5, 23, 25). The proliferation and differentiation of hematopoietic stem cells into different blood cell types require a robust upregulation of energy metabolism compared to less metabolically active BMAT in older rodents (26). Also, the need for energy caused by growing may be reflected in the BMGU. Taken together, high BMGU in younger rats may represent more active hematopoiesis and bone formation than low BMGU in skeletally mature and BMAT-enriched marrow of older rats.

At 32 weeks, sternal BMGU was significantly higher than humeral BMGU when all intervention groups were combined. Cline et al. showed in rats that cellular composition of humeral and sternal BM is different as sternal BM contains more hematopoietic tissue (27). Our data agrees with this finding, which is reflected by higher BMGU in the sternum than in humerus. The same phenomenon can be seen in humans when comparing vertebral and femoral BMGU (14, 15, 28). In humans, lumbar vertebral BM is more hematopoietically active while femoral BM cavity serves mainly as a specialized fat depot (14, 15, 28). Additionally, we found that humeral, but not sternal, BMGU with HFD exposure correlated inversely with whole-body fat percentage. It seems that higher BMGU tied to hematopoiesis in the axial skeleton is more independently regulated than BMGU in BM of long bones. Our data suggests that even small differences in BM cellularity can result in significant differences in BM metabolism and GU.

Despite differences in body weight, whole-body fat percentage, and glycemic status we did not observe any statistically significant differences in BMGU among intervention groups in either humerus or sternum. However, on average it seemed that both switching to chow and running exercise (¹²HFD+¹²(Chow+E)) increased BMGU compared to ²⁴HFD group. This agrees with our previous findings in humans, where we found that femoral BMGU increased in response to exercise training similarly in both healthy and insulin-resistant participants after two weeks of exercise training (14), and in co-twins discordant for body weight after six months of regular exercise training (15). However, in our previous twin study we showed that exercise training induced an increase in BMGU even without a decrease in BM adiposity measured using CT radiodensity (15). According to our data, it seems that this same phenomenon can be seen in rats and changes in BMGU and adiposity can occur independent of each other. To our knowledge, there are no other studies on BMGU under euglycemic-hyperinsulinemic clamp; however, Suchacki et al. compared the effects of insulin and vehicle on BMGU in fasting condition without continuous glucose infusion (13).

One reason that could explain the similar BMGU and whole-body insulin sensitivity (M-value) among intervention groups, unlike our previous findings in humans, is that rats were anesthetized with isoflurane while humans were awake during imaging (14–16, 28). Isoflurane anesthesia has been shown to cause hyperglycemia in humans and in rats (29–31). The increased levels of plasma glucose are a consequence of impaired insulin secretion, impaired glucose clearance and increased glucose production (29, 31). Indeed, anesthesia is needed to immobilize rodents for imaging, which might have affected the reliability of the parameters related to glucose measurements reported in this study. We have previously shown in humans that BM metabolism can be affected by exercise training (14, 15, 28). However, in this study, no statistically significant changes in BMGU were found in rats. Also, the effect of aging in decreasing insulin-stimulated BMGU may concomitantly mask the effect of our intervention. To elucidate the effect of anesthesia and aging on insulin sensitivity measurements more studies are required.

Although interventions reduced whole-body fat percentage, neither diet nor exercise intervention reduced BM adiposity close to the age-matched control (²⁴Chow). Indeed, BMAT has been reported to be distinct from extramedullary adipose depots (1–4), which may explain why BMAT was relatively resistant to the interventions in our experimental model. For instance, the limited lipolytic capacity of BMAds compared to subcutaneous white adipocytes may partly explain the resistance of BMAT to remodel under external stimuli (2, 4). Further, BMAT is presented with higher expression of inflammation-related genes compared to other extramedullary adipose tissues under basal condition (11); this may prime adipogenesis upon early exposure to the HFD. In accordance with this, BM global gene expression showed a pro-adipogenic and pro-inflammatory profile with HFD (32). In their study, rats with exercise intervention, similar to our ¹²HFD+¹²(HFD+E) group, remained pro-adipogenic but not pro-inflammatory when compared to the sedentary HFD rats; this may explain the relatively stable morphology of BMAds observed in our study. However, specific changes in BMAds remain to be investigated.

Interestingly, HFD significantly increased the median size of BMAds in the proximal but not distal tibia, representing diet-specific response in regulated but not constitutive BMAds, similar to the work by Scheller et al. (5). The established heterogeneity may partly explain the regional differences observed in the response of BMAds in the tibia in our model.

Previously, we and others have shown that BMAT-enriched human BM is an insulin-sensitive depot (3, 14, 15, 28). In this study, we show that HFD increased the size of regulated BMAds in the proximal tibia which correlated inversely with insulin-stimulated BMGU in both humerus and sternum, suggesting that BMAT is not protected from obesity. Although more studies are needed to mechanistically link the size of BMAds and BMGU, research on white adipocytes shows the association between adipocyte hypertrophy and insulin resistance (33–35). In this study, the analyzed regions for BM adiposity (proximal and distal tibia) and BMGU (humerus and sternum) were different. The rats were positioned in the PET/CT scanner head first, which means their hind legs were outside the axial field of view of the scanner and thus not included in the analysis. However, BM of the long bones, such as tibia and humerus are considered to be similar in BM cellularity, BMAT volume and BMGU (8, 13, 27).

To the best of our knowledge, this is the first study to combine dynamic [¹⁸F]FDG-PET imaging during euglycemic-hyperinsulinemic clamp and morphological histology to assess the effects of diet and exercise on BMGU and BM adiposity. While many studies have shown marked reduction in BM adiposity with diet and exercise intervention, our study model showed a rather moderate effect. This may have been the result of the use of less energy-dense (42% calories from fat) diet and voluntary running for four days per week compared to others (6, 36–38). Furthermore, the difference in the duration of the interventions and the age of animals may affect the response of BMAT to the interventions. For instance, aging-related expansion of BMAT (5, 12, 23, 39) may have masked the effect of intervention in older animals. Further, it might be that the effect of the interventions on basic anthropometrics as well as BMAT and BM metabolism could have been more pronounced after a longer or more rigorous exercise intervention or with a diet containing more fat. Our study intervention was designed to reflect the feasibility in clinical practice and provide mechanistic insights that parallels to our previous human studies (14, 15).

In summary, high-fat diet increases whole-body fat percentage and the size of regulated BMAds in the proximal but not that of constitutive BMAds in the distal region of the tibia, suggesting that BMAT is not protected from the effects of obesity. While a 12-week dietary and intermittent running exercise intervention is effective in reducing whole-body fat percentage, it does not significantly affect bone marrow adiposity. The size of regulated BMAds correlate inversely with bone marrow glucose uptake in the humerus. Insulin-stimulated bone marrow glucose uptake is site-specific and decreases with age and the study interventions are not able to affect it. To reverse bone marrow adipocyte hypertrophy and induce favorable changes in bone marrow glucose uptake, a more intense intervention may be required.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The animal study was approved by State Provincial Office of Southern Finland. The study was conducted in accordance with the local legislation and institutional requirements.

RO: Data curation, Formal analysis, Methodology, Writing – original draft, Writing – review & editing. NW: Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review & editing. JH: Writing – review & editing, Data curation, Formal analysis, Methodology. AJ: Writing – review & editing, Investigation. JSH: Writing – review & editing, Conceptualization, Investigation. TN: Writing – review & editing, Investigation. NJ: Writing – review & editing, Formal analysis. OE: Writing – review & editing, Investigation. JR: Writing – review & editing, Investigation. EL: Writing – review & editing, Formal analysis. KKI: Conceptualization, Funding acquisition, Methodology, Project administration, Resources, Supervision, Writing – original draft, Writing – review & editing. JCH: Conceptualization, Funding acquisition, Methodology, Project administration, Resources, Supervision, Writing – original draft, Writing – review & editing.