The caveolin-cavin system plays a conserved and critical role in mechanoprotection of skeletal muscle

Abstract

Dysfunction of caveolae is involved in human muscle disease, although the underlying molecular mechanisms remain unclear. In this paper, we have functionally characterized mouse and zebrafish models of caveolae-associated muscle disease. Using electron tomography, we quantitatively defined the unique three-dimensional membrane architecture of the mature muscle surface. Caveolae occupied around 50% of the sarcolemmal area predominantly assembled into multilobed rosettes. These rosettes were preferentially disassembled in response to increased membrane tension. Caveola-deficient cavin-1(-/-) muscle fibers showed a striking loss of sarcolemmal organization, aberrant T-tubule structures, and increased sensitivity to membrane tension, which was rescued by muscle-specific Cavin-1 reexpression. In vivo imaging of live zebrafish embryos revealed that loss of muscle-specific Cavin-1 or expression of a dystrophy-associated Caveolin-3 mutant both led to sarcolemmal damage but only in response to vigorous muscle activity. Our findings define a conserved and critical role in mechanoprotection for the unique membrane architecture generated by the caveolin-cavin system.

Figure 1.

Loss of Cavin-1 in mice recapitulates the skeletal muscle phenotype observed in patients. (A) Hematoxylin and eosin staining of WT and cavin-1^−/− muscle sections. Arrows indicate centralized nuclei in cavin-1^−/− muscle sections. Bar, 50 µm. (B) WT muscle has 1.2 ± 0.1% central nuclei compared with 7.1 ± 1.1% central nuclei in cavin-1^−/− muscle (determined by counting number of central nuclei in 564 and 596 muscle fibers from three pairs of WT and cavin-1^−/− mice, respectively). No further changes in histology were observed in cavin-1^−/− mice up to 20 mo of age (not depicted). (C–E) WT and cavin-1^−/− mice were tested for hang time (C), endurance capacity (D), and maximum running speed (E). WT mice had a mean hang time of 3.1 ± 1.0 min compared with 0.3 ± 0.1 min in cavin-1^−/− mice. Use of the holding impulse (hang time × body weight) measurement to correct for effects of body mass on hang time yielded the same results (not depicted). Exercise endurance capacity, measured as total running distance before exhaustion, was 871.9 ± 20.6 m in WT mice and 788.4 ± 37.6 m in cavin-1^−/− mice. Maximum running speed was 20.3 ± 0.8 m/min in WT mice and 19.2 ± 0.7 m/min in cavin-1^−/− mice. Tests performed on 10 WT and 6 cavin-1^−/− mice. Error bars show means ± SEM. *, P ≤ 0.05; **, P ≤ 0.01; ns, not significant. (F–H) Ruthenium red–labeled isolated FDB fibers from WT (F and G) and cavin-1^−/− (H) mice show the convoluted surface of caveola-deficient cavin-1^−/− fibers, in contrast to the straight sarcolemma of WT fibers decorated by rosettes of uniform surface-connected caveolae. Insets show reticular networks characteristic of immature Cav3-positive structures (Parton et al., 1997). Bars: (F) 2 µm; (G and H, insets) 0.5 µm; (H, main image) 2 µm. (I) Quantitation of sarcolemmal features in WT and cavin-1^−/− fibers. Unit of muscle membrane (micrometers) per sarcomere is expressed as a proportion (percentage) occupied by sarcolemmal surface excluding caveolae (in WT) and surface-connected vacuoles (in cavin-1^−/−). Sarcolemmal surface occupies 41% and 37% in WT and cavin-1^−/− in muscle fibers, respectively (unshaded). Caveolae occupy 59% of the sarcolemma in WT muscle fibers, whereas vacuoles occupy 63% of the sarcolemma in cavin-1^−/− muscle fibers (shaded gray). Note that stereological measurements have not been corrected for overprojection effects (see Materials and methods).

Figure 2.

Loss of Cavin-1 in muscle is associated with large vacuoles and a dilated T-tubule system. (A and B) Ruthenium red labeling of WT muscle fibers highlighting the muscle fiber surface, T tubules (inset in A represents enlargement of boxed area) and caveolar rosettes (inset in B represents enlargement of boxed area). (C–E) Ruthenium red labeling of cavin-1^−/− muscle fibers across the surface and within T tubules (arrows, C and D) and large vacuoles (arrows, E). Bars: (A and B, main images) 2 µm; (A and B, insets) 200 nm; (C–E) 1 µm.

Figure 3.

Ultrastructural morphological abnormalities do not affect the relative surface area of cavin-1^−/− muscle fibers. (A–D) Single tomogram image from a ruthenium red–labeled WT muscle fiber showing T-tubule connection to the surface via caveolae (A). Tomographic slices (B–D) represent an enlargement of boxed area in A. See Video 2. (E–G) High magnification 3D images of a WT muscle fiber showing T-tubule connection to the surface via caveolae. See Video 3. (H) 3D surface-rendered reconstruction showing T-tubules connecting to the muscle fiber surface via caveolar rosettes. (I–L) Single tomogram image of a ruthenium red–labeled cavin-1^−/− muscle fiber showing dilated T-tubules connecting to the surface via vacuoles (I). Tomographic slices (J–L) represent an enlargement of boxed area in I. See Videos 4 and 5. (M) Tomographic slice of a ruthenium red–labeled cavin-1^−/− muscle fiber. (N) Surface-rendered reconstruction highlights honeycomb structure of T tubule–connected reticulated networks as previously shown in differentiating muscle cells (Parton et al., 1997) and human CAV3 muscle (Minetti et al., 2002). Reticulated networks are connected to the surface (S) via vacuoles (V). (N1–N3) Tomographic slices of ruthenium red–labeled cavin-1^−/− muscle fiber area highlighted in N. See also Fig. S3 (A and B). (O) 3D view of surface-rendered reconstruction of area highlighted in N showing reticulated networks observed in cavin-1^−/− muscle (green). Pink highlights connections to the plasma membrane. Bars: (A, I, and M) 500 nm; (B, E, and J) 100 nm; (H, N, N1–N3, and O) 200 nm.

Figure 4.

Loss of Cavin-1 affects T-tubule and sarcolemmal organization in muscle. (A) Colabeling for TRPC1 and DHPR and labeling for Dysf and Bin1 at the subsarcolemmal surface of WT and cavin-1^−/− muscle fibers. (top) Insets show a typical T-tubule staining pattern for TRPC1, DHPR, Dysf, and Bin1 in WT muscle fibers. (bottom) Dense areas of staining for TRPC1, DHPR, Dysf, and Bin1 (arrows, insets) were observed within cavin-1^−/− muscle fibers. Insets show a higher magnification. (B) Cavin-4/Cav3 colabeling across the surface of WT and cavin-1^−/− muscle fibers. Increasing the signal intensity of Cavin-4 staining revealed that Cavin-4/Cav3 colocalization was lost in cavin-1^−/− muscle fibers (insets represent boxed areas). Arrows indicate areas of dense Cav3 staining. (C) Duolink PLA in WT and cavin-1^−/− muscle fibers using anti–Cavin-4 and anti-Cav3 antibodies with DAPI counterstain. Negative control (nc) was performed on WT muscle fibers where one primary antibody was omitted. (D) Cavin-4 staining in methanol-fixed WT and cavin-1^−/− muscle fibers with DAPI counterstain. Inset represents overlay of Cavin-4/DAPI labeling within nucleus in boxed areas. Muscle fibers are highlighted with dashed lines. Bars: (A) 20 µm; (B–D) 10 µm; (A, B, and D, insets) 5 µm.

Figure 5.

Preferential disassembly of caveolar rosettes in response to increased membrane tension. (A and B) Immunofluorescence of Cavin-1/Cav3 (A) and Cavin-4 (B) in WT muscle fibers after 15 min in hypo-osmotic (HYPO) media (insets represent brightfield image of membrane bleb). (C) Duolink PLA of Cavin-4/Cav3 interaction in WT muscle fibers after 15-min incubation in hypo-osmotic media (inset, brightfield image of membrane bleb). Dashes outline the membrane bleb. (D) EM analysis of a ruthenium red–labeled WT muscle fiber after 15 min in hypo-osmotic media (inset, higher magnification of bleb area indicated by asterisks). (E) Relative caveolae density of WT fibers in hypo-osmotic media was 39.6 ± 16.3% (means ± SD; compared with WT fibers incubated in iso-osmotic [ISO] media). (F) Ruthenium red–labeled WT muscle fibers in iso-osmotic or hypo-osmotic media. Arrows indicate caveolae beneath the sarcolemma. Several single caveolae are observed in hypo-osmotic–treated muscle fibers (inset). (G) Quantitation of number of caveolae associated with caveolar rosettes (caveolae/rosette) after incubation in iso-osmotic or hypo-osmotic media: 43 ± 7%, 17 ± 1%, 30 ± 4%, 5 ± 6%, and 4 ± 2% of caveolae from muscle fibers in iso-osmotic media had 1–5 caveolar rosettes, respectively; 86 ± 9%, 10 ± 9%, 3 ± 1%, 1 ± 2%, and 0 ± 0% of caveolae from muscle fibers in hypo-osmotic media had 1–5 caveolar rosettes, respectively. Error bars show means ± SD. Quantitation was performed on >440 caveolae from three different muscle fibers for both iso-osmotic and hypo-osmotic treatments. (H) 3D view of surface-rendered reconstructions of caveolar rosettes (yellow) in WT fibers incubated in iso-osmotic or hypo-osmotic media. See Videos 6 and 7. Bars: (A–C) 10 µm; (D and F) 2.5 µm; (D [inset] and H) 200 nm. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

Figure 6.

Exogenous Cavin1-GFP expression in cavin-1^−/− muscle fibers rescues the null phenotype. (A) Western analysis of cavin-1^−/− muscle fibers expressing GFP reporter or Cavin1-GFP using anti-GFP and anti-Cav3 antibodies. β-Actin demonstrates protein loading; 50 µg protein lysate was loaded. (B–K) Cavin-4 (B–F) and Cav3 (G–K) immunostaining in cavin-1^−/− muscle fibers expressing GFP reporter (B and G) or Cavin1-GFP (D and I). Insets represent enlargement of boxed areas. (L–O) Duolink PLA using anti–Cavin-4 and anti-Cav3 antibodies in cavin-1^−/− muscle fibers expressing GFP reporter (L) or Cavin1-GFP (N). Muscle fibers are highlighted with dashed lines. (P) PLA signal in cavin-1^−/− muscle fibers was 0.0007 ± 0.0007 dots/µm² and 0.014 ± 0.007 dots/µm² in fibers expressing GFP reporter or Cavin1-GFP, respectively. Quantitation performed on 15 GFP reporter and 9 Cavin1-GFP–expressing muscle fibers from AAV injections of two cavin-1^−/− mice. (Q) Ruthenium red–labeled muscle fibers expressing GFP reporter or Cavin1-GFP. Arrows indicate single caveolae (middle) and rosettes (right). (R) Mean percentage of cell lysis was 36.3 ± 6.9% and 11.0 ± 3.4% in cavin-1^−/− muscle fibers expressing GFP reporter or Cavin1-GFP, respectively, and is represented as a column graph. Quantitation was performed on 85 and 99 muscle fibers expressing GFP reporter or Cavin1-GFP, respectively, from six cavin-1^−/− mice. Dot plots represent percentage of cell lysis observed for individual mice; note that the percentage of cell lysis is consistently higher in the absence of Cavin-1. Error bars are means ± SEM. *, P ≤ 0.05; **, P ≤ 0.01. Bars: (B–O) 10 µm; (Q, left and middle images) 1 µm; (Q, right image) 200 nm.

Figure 7.

Caveolae are essential for maintaining muscle integrity in the zebrafish. (A) Temporal expression of cavin-1a mRNA in 1.3–72-hpf zebrafish embryos. (B) Expression pattern of cavin-1a in zebrafish embryos using whole-mount mRNA in situ hybridization. Dorsal view of a 12-somite (12S) embryo shows faint labeling for cavin-1a (arrowheads). At 48 hpf, cavin-1a expression is observed exclusively within the zebrafish myotomes, shown here in dorsal (top right) and lateral (bottom left) view; anterior to left in both images. Note the lack of notochord labeling in the dorsal view. Image in bottom right represents magnification of boxed area. (C) WT and tp53^zdf1 zebrafish embryos were injected with control or cavin1a MO (CtrMO and cavin1aMO, respectively) and abnormal morphants (classified as mild or severe) imaged at 72 hpf. Arrowheads indicate cardiac edema. (D) Ruthenium red–labeled isolated muscle fibers from control MO and cavin1aMO embryos. Arrows indicate caveolae (left image) or endosomes (middle and right images). Inset represents boxed area. (E) Relative caveolae density of muscle fibers from cavin1aMO embryos was 1.4 ± 2.0% (means ± SD), compared with muscle fibers from control MO embryos. Quantitation performed on four control MO and six cavin1a MO muscle fibers. (F) EBD uptake in 96 hpf Tg(actb2:EGFP-CAAX)^pc10 embryos expressing EGFP-CAAX injected with control MO or cavin1a MO and incubated in 3% MC. Inset shows higher magnification of EGFP-CAAX/EBD-positive muscle fibers. (G) 0.0 ± 0.0% of control MO and cavin1aMO embryos were EBD positive after incubation in E3 (means ± SEM; 28 control MO and 47 cavin1a MO embryos from three separate microinjections). 0.8 ± 0.8% of control MO and 17.0 ± 7.0% of cavin1a MO embryos were EBD positive after incubation in 3% MC (means ± SEM; 105 control MO and 111 cavin1a MO embryos from eight separate microinjections). (H) EBD uptake in 96 hpf embryos expressing Cav3-WT-GFP (WT) or Cav3-R26Q-GFP (R26Q) after incubation in 3% MC. Inset shows higher magnification of GFP/EBD-positive muscle fibers. (I) 0.8 ± 0.8% of WT and 2.1 ± 2.1% R26Q embryos were EBD positive after incubation in E3 (means ± SEM; 71 WT and 33 R26Q embryos from five and four clutches, respectively). 7.1 ± 2.6% of WT and 31.5 ± 5.9% R26Q embryos were EBD positive after incubation in 3% MC (130 WT and 126 R26Q embryos from seven and six clutches, respectively). Uptake was performed in two separate founder lines for both WT and R26Q to ensure that phenotypes observed were not caused by Tol2 integration sites. *, P ≤ 0.05; **, P ≤ 0.01. ns, not significant. Bars: (B and C) 200 µm; (D, main images) 1 µm; (D, inset) 200 nm; (F and H, main images and insets) 50 µm.