Region-specific variation in the properties of skeletal adipocytes reveals regulated and constitutive marrow adipose tissues

Abstract

Marrow adipose tissue (MAT) accumulates in diverse clinical conditions but remains poorly understood. Here we show region-specific variation in MAT adipocyte development, regulation, size, lipid composition, gene expression and genetic determinants. Early MAT formation in mice is conserved, whereas later development is strain dependent. Proximal, but not distal tibial, MAT is lost with 21-day cold exposure. Rat MAT adipocytes from distal sites have an increased proportion of monounsaturated fatty acids and expression of Scd1/Scd2, Cebpa and Cebpb. Humans also have increased distal marrow fat unsaturation. We define proximal 'regulated' MAT (rMAT) as single adipocytes interspersed with active haematopoiesis, whereas distal 'constitutive' MAT (cMAT) has low haematopoiesis, contains larger adipocytes, develops earlier and remains preserved upon systemic challenges. Loss of rMAT occurs in mice with congenital generalized lipodystrophy type 4, whereas both rMAT and cMAT are preserved in mice with congenital generalized lipodystrophy type 3. Consideration of these MAT subpopulations may be important for future studies linking MAT to bone biology, haematopoiesis and whole-body metabolism.

Figure 1. Quantification of MAT development in C57BL/6J and C3H/HeJ mice from 1 to 56 weeks of age.

(a) Osmium-stained tibiae were scanned by μCT and were reconstructed with the decalcified bone overlaid. Representative images of the data are presented in (b). Marrow fat is dark grey and bone is light grey. (b) Region-specific quantification of MAT volume (biological replicate N=5 (1 week old), 7 (4 weeks old), 9 (C3H 12 weeks old) and 11 (B6 12- and 56 weeks old)). Regions as defined in a include the proximal epiphysis (Prox Epi), the growth plate to the tibia/fibula (Tib/Fib) junction (GP to T/F J) and the tibia/fibula junction to the end of the bone (T/F J to end). ^aTwo-tailed t-test, P<0.05 for total tibial MAT volume compared between strains at a given age. (c) Representative histology of caudal vertebrae (biological replicate N = 5), × 10 objective (scale bars, 200 μm). All graphs represent mean ± s.d.

Figure 2. Trabecular and cortical development in C57BL/6J and C3H/HeJ mice at 12 and 56 weeks of age.

(a,b) Representative images of the proximal tibial metaphysis both before decalcification and after osmium staining of the data presented in c. Marrow fat is in white. (c) Quantification of trabecular parameters and MAT volume in the proximal tibial metaphysis (biological replicate N=9 (C3H 12 weeks old) and 11 (B6 12- and 56 weeks old)). (d,e) Representative images of the mid-tibial diaphysis both before decalcification and after osmium staining of the data presented in panel f. Marrow fat is in white. (f) Quantification of cortical parameters (biological replicate N=9 (C3H 12 weeks old) and 11 (B6 12- and 56 weeks old)). Scale bars=500 μm. *Two-tailed t-test, P<0.05. All values represent mean±s.d.

Figure 3. Differential loss of MAT with cold exposure.

(a) Representative osmium-stained tibiae scanned with μCT of the data presented in b and c. Marrow fat in dark grey, decalcified bone overlaid in light grey. (b) Region-specific quantification of tibial MAT (as defined in Fig. 1a) from 12-week-old mice normalized to marrow volume (biological replicate N=11 (4 °C) and 11 (22 °C)). (c) Region-specific quantification of tibial MAT from 56-week-old mice normalized to marrow volume (biological replicate N=9 (4 °C) and 11 (22 °C)). (d) Adipocyte size distribution from the proximal tibial metaphysis (proximal) or (e) distal diaphysis below the tibia/fibula junction (distal) as measured by nanoCT in the 12-week-old mice (biological replicate N = 5). Histogram bin size 250. (f) Representative histology, based on quantification in d, of osmium-stained samples at the proximal tibia shows a decrease in adipocyte size. Scale bars, 50 μm. (g) Estimation of region-specific adipocyte number was performed by dividing the total adipocyte volume (from μCT) by the average adipocyte volume (nanoCT) in the proximal tibia (growth plate to tibia/fibula junction) and the distal tibia (tibia/fibula junction to the distal end). *(b,c,g) Two-tailed t-test, (d,e) two-way analysis of variance with Sidak’s multiple comparisons test, P<0.05. RT, room temperature. All graphs represent mean±s.d.

Figure 4. Quantification of rMAT and cMAT adipocyte size.

Adipocyte size quantification from (a) rMAT in the proximal tibia and cMAT in the distal tibia of 12-week-old C3H mice (biological replicate N=5), (b) rMAT in the proximal tibia and cMAT in the caudal vertebrae of 16-week-old male Sprague–Dawley rats (biological replicate N=12), and (c) rMAT in the mid-tibia and cMAT in the caudal vertebrae of 19-week-old female Sprague–Dawley rats (biological replicate N=5 rMAT and 6 cMAT). *Two-way analysis of variance with Sidak’s multiple comparisons test, P<0.05. All graphs represent mean±s.d.

Figure 5. Region-specific lipid saturation of human and rat MAT.

(a) Marrow unsaturation at four sites in the human leg was compared using ¹H-MRS. Marrow at the metaphysis of the distal tibia (T) had a higher unsaturation index than marrow in the proximal femur (F) (biological replicate N=5). Mean±s.e. (b) Adipocytes were isolated from four regions of the rat skeleton. Red outlines indicate rMAT sites including femur/proximal (Prox) tibia and lumbar vertebrae. Grey and black outlines indicate cMAT sites including distal tibia and caudal vertebrae. Histology of intact rat rMAT and cMAT before adipocyte isolation representative of the animals from Fig. 4b (biological replicate N=12). Objective × 40, scale bars, 50 μm. (c) Principal components analysis of normalized fatty acids from three independent experiments (23 fatty acids and 44 unique biological samples). Raw data presented in Supplementary Data 1. Visceral WAT (vWAT) includes 5 gonadal and 3 perirenal; subcutaneous WAT (scWAT) includes 12 inguinal; rMAT includes 3 lumbar vertebrae and 6 femur/proximal tibia; cMAT includes 3 distal tibia and 12 caudal vertebrae (samples were derived from 13 unique animals). Dashed line = 95% confidence interval. (d) Proportion of fatty acids with one or more double bonds relative to total lipid in adipocytes from perirenal visceral WAT (vWAT), inguinal scWAT, rMAT from femur/proximal tibia (rMAT T/F), rMAT from lumbar vertebrae (rMAT Vert), cMAT from distal tibia (cMAT tibia) and cMAT from tail vertebrae (cMAT Vert). Representative data presented as mean±s.d. (as presented, biological replicate N=3). Experiment repeated with similar results in three animal cohorts with a total of 44 samples from 13 rats as outlined in Supplementary Data 1. *One-way analysis of variance with Tukey’s multiple comparisons test, P<0.05.

Figure 6. Gene expression of desaturases in isolated adipocytes.

(a) Proportion of C16:1n7-palmitoleate relative to C16:0-palmitate. (b) The proportion of C18:1n9-oleate relative to C18:0-stearate. Representative data presented as mean±s.d. (as presented, biological replicate N=3). Repeated with similar results in three animal cohorts with samples from 13 total rats as detailed in Supplementary Data 1. Transcript expression in isolated constitutive MAT (cMAT) and subcutaneous WAT (scWAT) adipocytes normalized to scWAT from (c) 16-week-old male Sprague–Dawley rats (biological replicate N=6 cMAT (two animals pooled per sample) and N=12 scWAT) and (d) 8-month-old female Sprague–Dawley rats (biological replicate N=3 cMAT (two animals pooled per sample) and N=5 scWAT). Presented as mean±s.d. *(a,b) One-way analysis of variance with Tukey’s multiple comparisons test, (c,d) two-tailed t-test, P<0.05.

Figure 7. Gene expression of transcription factors in isolated adipocytes.

Transcript expression in isolated adipocytes from subcutaneous WAT (scWAT), constitutive MAT (cMAT) and/or regulated MAT (rMAT) normalized to scWAT from (a) 16-week-old male Sprague–Dawley rats (biological replicate N=6 cMAT (two animals pooled per sample) and 12 scWAT), (b) 8-month-old female Sprague–Dawley rats (biological replicate N=3 cMAT (two animals pooled per sample) and 5 scWAT), and (c) 16-week-old male Sprague–Dawley rats (biological replicate N=5, four animals pooled per sample). Presented as mean±s.d. *(a,b) Two-tailed t-test, (c) one-way analysis of variance, P<0.05.

Figure 8. Differential loss of MAT in mice with knockout of Cav1 or Ptrf.

(a) Representative osmium-stained tibiae scanned with μCT based on data in b and c. Marrow fat in dark grey, decalcified bone overlaid in light grey. (b,c) Region-specific quantification of tibial MAT volume by μCT (biological replicate N=6–9 as indicated on the graph). Box plot centre line represents median, box extends from the 25th to 75th percentile, whiskers indicate range. (d) Representative histology of caudal vertebrae based on data in e, × 10 objective (scale bars, 200 μm). (e) Adipocyte size distribution of the caudal marrow adipocytes as measured by histology (biological replicate N=5). Histogram bin size 250. Presented as mean±s.d. *(b) Non-parametric Mann–Whitney test, (c) two-tailed t-Test, (e) two-way analysis of variance with Sidak’s multiple comparisons test, P<0.05. KO, knockout; WT, wild type.

Figure 9. Trabecular morphology versus MAT in the tibia and L4 vertebrae of Ptrf KO mice.

(a,b) Representative images of the proximal tibial metaphysis both before decalcification and after osmium staining based on data in c. Marrow fat is in white. Scale bars, 500 μm. (c) Quantification of trabecular parameters and MAT volume in the proximal tibial metaphysis (biological replicate N=5 (female KO), 6 (female WT), 7 (male KO) and 9 (male WT)). (d) Quantification of trabecular parameters in the vertebral body of L4 (biological replicate N=5 (WT) and 4 (KO)). KO, knockout; WT, wild type. All values represent mean±s.d. *Two-tailed t-test, P<0.05. ^aNon-parametric Mann–Whitney test, P<0.05.

Figure 10. rMAT versus cMAT summary.

In the human and mouse tibia, cMAT is present in the distal portion of the bone. The marrow shifts to red towards the proximal tibia, occurring near the tibia/fibula junction in the rodent and in the proximal tibial metaphysis or femur in the human. The red marrow contains rMAT adipocytes. In some cases, especially in larger species such as humans, the histological patterns that correspond to rMAT and cMAT adipocytes may be present in the same region. The bones have been shaded in orange to indicate this possibility. cMAT is the first to develop and histologically appears as sheets of confluent adipocytes that are relatively devoid of haematopoiesis. Isolated cMAT adipocytes form shortly after birth, have an increased proportion of unsaturated fatty acids and are larger in size. rMAT develops throughout life and is histologically defined as single cells interspersed with areas of active haematopoiesis. Isolated rMAT adipocytes have a lipid saturation profile that is similar to WAT adipocytes and are more saturated, and smaller in size, than cMAT. These cells are negatively regulated by 21-day cold exposure. They also fail to form in mice with genetic knockout of Ptrf, but not Cav1. NC, no change. Scale bar, 50 μm.