Superstable lipid vacuoles endow cartilage with its shape and biomechanics

Abstract

Conventionally, the size, shape, and biomechanics of cartilages are determined by their voluminous extracellular matrix. By contrast, we found that multiple murine cartilages consist of lipid-filled cells called lipochondrocytes. Despite resembling adipocytes, lipochondrocytes were molecularly distinct and produced lipids exclusively through de novo lipogenesis. Consequently, lipochondrocytes grew uniform lipid droplets that resisted systemic lipid surges and did not enlarge upon obesity. Lipochondrocytes also lacked lipid mobilization factors, which enabled exceptional vacuole stability and protected cartilage from shrinking upon starvation. Lipid droplets modulated lipocartilage biomechanics by decreasing the tissue's stiffness, strength, and resilience. Lipochondrocytes were found in multiple mammals, including humans, but not in nonmammalian tetrapods. Thus, analogous to bubble wrap, superstable lipid vacuoles confer skeletal tissue with cartilage-like properties without "packing foam-like" extracellular matrix.

Fig. 1:. Lipochondrocytes are present in multiple cartilages and impart biomechanical properties.

(A) Schematic drawing of mouse cartilages color-coded by embryonic origin. LC-containing neural crest (NC)-derived lipocartilage shown in green. (B-H) OilRedO staining and lineage tracing in Wnt1-Cre2;mT/mG mice show lipid-laden LCs in NC-derived nasal (B-D), epiglottic and thyroid cartilage (E-H), except for the posterior end (insets in F, G). See Fig. S2. (I-L) Ear lipocartilage consists of OilRedO⁺ and BODIPY⁺ LCs and contains fenestrae occupied by adipocytes. (M) Biomechanical analyses of mouse ear lipocartilage (Ear) compared to WAT. Despite morphological similarities, lipocartilage displays significantly higher stiffness (Young’s modulus; left; 3.25±0.38 MPa; Welch’s t test, p=0.0004), ultimate tensile strength (UTS; center; 1.30±0.24 MPa; Welch’s t test, p=0.005), than WAT, but lower strain at failure (SAF; right; 0.60±0.10, p=0.0273). (N) Biomechanical analyses of native rat ear lipocartilage (control) and delipified tissue. Compared to control (green), stiffness (left; 3.27±0.52 vs. 5.79±0.99 MPa; Welch’s t test, p=0.0004), UTS (center; 1.91±0.22 vs. 4.53±0.92 MPa; Welch’s t test, p=0.0259), and resilience (right; 0.38±0.06 vs. 1.77±0.48 MJm⁻³; Welch’s t test, p=0.0228) increase significantly following delipification (gray). (O-R) Clonogenic analysis in Col2a1-CreER^T;R26R mice shows lacZ-labeled LCs form discrete clones that are stable in size for six months (Kolmogorov-Smirnov test, p=0.506). Scale bars: (H), 25 μm; (L, P, Q), 100 μm; (B, C, F), 200 μm; (G), 250 μm; (O), 500 μm; (E, I), 1 mm.

Fig. 2:. Lipochondrocytes have unique multiomic profile.

(A) Lipidomic profile of ear lipocartilage (blue) is distinct from that of inguinal WAT (yellow) on PCA analysis. (B) Lipocartilage and WAT are differentially enriched for distinct lipid classes. See Table S1 for the definition of lipid classes. Prominently, lipocartilage is enriched for free fatty acids (FFA) and fatty acid hydroxyl-fatty acids (FAHFA), while WAT for triacylglycerols (TAG) and diacylglycerols (DAG). (C) Compared to WAT, lipocartilage contains substantially more lipids with saturated FFA tails (green). (D-I) Proteomic (D, G) and immunostaining analyses (E-I) show marked differences in structural and lipogenesis protein content between ear lipocartilage and WAT. Prominently, lipocartilage is enriched for COL8A1 and MYOC (D), which have distinct localization patterns (E, F). SG: sebaceous gland, HF: hair follicle. (J, K) PCA and heatmap analyses show that transcriptional signature of ear LCs (green) is distinct from that in adipocytes (blue) and rib hyaline chondrocytes (red). See Table S3. (L) All three cell types show distinct enrichments for Gene Ontology terms. See Table S4. (M-O) List of selected genes differentially upregulated in LCs (M), rib chondrocytes (N) or adipocytes (O) are shown. See Table S3. Scale bars: (E, F, H, I), 20 μm.

Fig. 3:. Lipocartilage activates and depends on de novo lipogenesis.

(A) Schematic summary of the key stages and cellular events in ear lipocartilage development. RNA-seq time points are defined and annotated. (B, C) PCA and heatmap analyses show transcriptional changes in developing ear LCs at selected postnatal time points. See Table S5. (D) Relative abundance of genes across clusters T1-T4 from (C) within selected gene categories. (E-H) Stimulated Raman scattering (SRS) analysis reveals that glucose-derived deuterium (aka heavy hydrogen atom) incorporates into lipid vacuoles of developing ear LCs. Color-coded images of ear cartilage reveal total lipid within carbon-hydrogen bond (C-H) vibration range (E) and de novo lipogenesis-derived lipid within carbon-deuterium bond (C-D) vibration range (F). Raman scan on LC lipid vacuoles cultured in the presence of deuterated glucose (G-d7) shows C-H stretch peak, common to lipids, as well as a C-D stretch peak that overlaps with the C-D peak of pure G-d7 (G, red region). Overlay of (C) and (D) is shown in (H). (I-N) Ears become smaller in mice topically treated with de novo lipogenesis inhibitors PF-05175157 (PF), ND-646 (ND) and C75 as compared to DMSO control (I). C75-treated ears show prominent decrease in BODIPY⁺ lipid vacuoles (J). Representative mice and dissected one-month-old ears are shown in (K-N). Scale bars: (E, F, H), 25 μm; (J), 50 μm; (K-N, bottom), 2 mm.

Fig. 4:. Lipocartilage maintains super-stable lipid vacuoles.

(A-D’) Ears (A-D) and individual LCs (A’-D’) do not significantly change in size in mice on a high fat diet (HFD), obese db/db mice or mice on restricted calorie intake (RCI) as compared to regular chow (RC) mice. Representative images are shown. In contrast to ear size (Welch’s ANOVA, p=0.0966) (E) and LC size (one-way ANOVA, p=0.2419) (F), body weight (Welch’s ANOVA, p=0.0002) (G) as well tissue adiposity (see Fig. S16) significantly increase in HFD (39.43±1.10 g; Dunnett’s T3, p=0.0005) and db/db animals (49.36±2.51 g; Dunnett’s T3, p=0.0396) and decrease in RCI conditions (23.38±0.79 g; Dunnett’s T3, p= 0.0100) relative to RC mice (30.36±0.85 g). (H-I) After treating mice with C1-BODIPY-C12 FFA for either one day (H) or eight days (I), adipocytes but not LCs incorporate fluorescent FFA. Experimental timelines are shown at the top. Scale bars: (A’-D’, H, I), 100 μm; (A-D), 2 mm.

Fig. 5:. Human head and neck cartilage accumulates lipid vacuoles.

(A) Alcian blue staining of human fetal ear cartilage at gestation week (GW) 20–21. (B) Immunostaining patterns of COL8A1 and PLIN1 in GW20–21 human ear. Adipocyte marker PLIN1 is confined to dermal adipocytes and is absent in chondrocytes. (C-D) OilRedO staining pattern in GW20–21 human ear. Numerous lipid vacuoles are present in chondrocytes. (E-H) Electron micrographs of GW20–21 human thyroid cartilage (E), epiglottis (F), ear cartilage (G), and nasal cartilage fold (H) reveal chondrocytes with large lipid vacuoles. Nuclei and lipid vacuoles are pseudo-colored purple and red, respectively. (I) Alcian blue staining of hESC-derived cartilage pellet at in vitro culture day 65. (J-K) OilRedO staining of day 65 pellets shows numerous lipid vacuoles in chondrocytes. (L) Label-free imaging of day 65 cartilage pellets reveals lipid vacuoles (SRS signal, red) and fibrillar collagen (second harmonics generation signal, blue). (M-N) PCA and heatmap analyses show dynamic gene expression changes in hESC-derived cartilage pellet across three time points, days 21, 30 and 45. 4,715 genes are differentially expressed across these time points (Table S8). (O) List of selected genes within ECM, signaling and transcriptional factor (TF) categories and their expression levels in hESC-derived cartilage pellets at days 21 and 45. (P) Relative expression values of hallmark adipogenesis and lipogenesis genes in hESC-derived pellets across days 21, 30 and 45 measured on qRT-PCR. Scale bars: (E-H), 2 μm; (D), 10 μm; (K), 15 μm; (L), 25 μm; (J), 50 μm; (A-C, I), 100 μm.

Fig. 6:. Lipocartilage is a prevalent feature of mammals.

(A) Phylogenetic tree of extant mammalian clades, with branches representing monotremes (black), marsupials (yellow), and placentals, further separated by order (different colors indicate super-orders). Table on the right provides information on the number of species, for which ear cartilage was analyzed within each taxon and if OilRedO⁺ LCs were detected (see Table S9). (B-C) Representative examples of ear lipocartilage in rodents: (B) Acomys cahirinus (Cairo spiny mouse), (C) Jaculus jaculus (Lesser egyptian jerboa). (D-G) Representative examples of ear lipocartilage in non-rodent mammals: (D) Dasyuroides byrnei (Kowari; marsupial), (E) Petaurus norfolcensis (Squirrel glider; marsupial), (F) Glossophaga soricina (Pallas’s long-tongued bat; order Chiroptera) and (G) Elephantulus myurus (Eastern rock elephant shrew; super-order Afrotheria). Cartilage fenestrae and fenestra-associated adipocytes are evident in many species. Squirrel glider ear cartilage on (E) shows fenestra-associated hair follicles (HFs). Pallas’s long-tongued bat ear cartilage on (F) shows complex distribution of LCs, with sparse LCs along the medial side and ridge-like LC “stacks” along the lateral side. Scale bars: (A), 20 million years; (D, G), 1 mm; (B, C, E, G), 2 mm.