Prominent actin fiber arrays in Drosophila tendon cells represent architectural elements different from stress fibers

Abstract

Tendon cells are specialized cells of the insect epidermis that connect basally attached muscle tips to the cuticle on their apical surface via prominent arrays of microtubules. Tendon cells of Drosophila have become a useful genetic model system to address questions with relevance to cell and developmental biology. Here, we use light, confocal, and electron microscopy to present a refined model of the subcellular organization of tendon cells. We show that prominent arrays of F-actin exist in tendon cells that fully overlap with the microtubule arrays, and that type II myosin accumulates in the same area. The F-actin arrays in tendon cells seem to represent a new kind of actin structure, clearly distinct from stress fibers. They are highly resistant to F-actin-destabilizing drugs, to the application of myosin blockers, and to loss of integrin, Rho1, or mechanical force. They seem to represent an important architectural element of tendon cells, because they maintain a connection between apical and basal surfaces even when microtubule arrays of tendon cells are dysfunctional. Features reported here and elsewhere for tendon cells are reminiscent of the structural and molecular features of support cells in the inner ear of vertebrates, and they might have potential translational value.

Figure 1.

Resolving the apico-basal axis of tendon cells in the confocal microscope. Late larval tendon cells are shown in various views, as indicated bottom right in all images (anterior to the left): vertical views are presented in frontal (f) or sagittal (s) orientation, horizontal views are presented in either of three focal plains (b, basal; i, intermediate; a, apical), as illustrated by red arrows in the cartoon in B; alternatively horizontal views are shown as maximal projections (p), or as partial projections (b/i, basal plus intermediate; i/a, intermediate plus apical). (A) Direct muscle attachment (asterisk) shown as a maximal projection in horizontal view; the cell surface marker mCD8::GFP (CD8) demonstrates the tendon cell's shape and its close association with the phalloidin-stained muscle (Phal; white dot); Dlg illustrates the cell perimeter (curved white arrow) at the level of intraepidermal septate junctions. (B) Model of tendon cell in sagittal and two horizontal views explaining its overall organization and introducing symbols used throughout this paper: asterisk, tendon cell; white dot, muscle; white/black arrow, distal/proximal end of MTJ; white/black arrow heads, basal/apical end of cytoskeletal arrays; used abbreviations: bm, basement membrane; ca, cytoskeletal array (tubulin in green, F-actin in red); cu, cuticle; ep, normal epidermal cell; mtj, myotendinous junction; sj, septate junction. All other images show different views (bottom right) of phalloidin-labeled muscles colabeled with tendon cell-specific tubulin::GFP (Tub; C¹–C⁴), GFP-tagged Neurexin (Nrx, labels septate junctions; D¹–D⁴), or Viking::GFP (Vkg, labels basement membrane; E¹–E⁴); symbols as in B. Bar, 22 μm (A) and 25 μm (C–E).

Figure 2.

Properties of microtubule arrays in tendon cells. Horizontal views of muscles (white dots) and tendon cells (asterisks) taken in intermediate (i) or basal (b) focal planes, according to Figure 1B. Usual networks of microtubules at the surface of muscles and tendon cells (A¹) are severely reduced or abolished upon nocodazole treatment (B¹); curved arrows point at nerves reaching into the image. In control animals, cytoskeletal arrays are labeled with tubulin::GFP (magenta), whereas anti-tubulin staining mostly fails to penetrate deep into tendon cells (A²); in nocodazole-treated animals the cytoskeletal belts are perfectly labeled by both anti-tubulin and tubulin::GFP (B²). Cytoskeletal belts can also be labeled with other anti-tubulin antibodies (E; Glu, anti-glu-tubulin; Tyr, anti-tyrosinated tubulin) or tubulin-associated proteins: Short stop::GFP (ShotGFP; F), antibody staining against C-terminal Short stop (ShotC; G) and EB1::GFP (EB1GFP; H). Bar is 10 μm (A and B) and 17 μm (E–H).

Figure 3.

Refining localization studies via enzymatic muscle detachment. (A¹–E⁵) Different views of control (con) or trypsin-treated (Tryp) direct muscle attachments; planes of view (bottom right) and symbols as explained in Figure 1B. Specimens are labeled with phalloidin (Phal) together with either tendon cell-specific tubulin::GFP (Tub) or GFP-tagged endogenous Ilk. The organization of tubulin into a cytoskeletal belt (between arrow heads in A²) and dashed/stippled fields (A³) is maintained in tendon cells after muscle detachment (B² and B³), i.e., the muscle leaves behind its footprint in the orphan tendon cell (B⁴). Ilk::GFP is restricted to the immediate contact zone of muscle and tendon cells, decorating only the basal end of cytoskeletal arrays (C¹). On muscle detachment, Ilk::GFP is found at the muscle tip (D), but also on the basal surface of the cytoskeletal arrays (E¹–E⁵). Electron micrographs taken from collagenase IV-treated (Col IV) embryos show a detached muscle (F), an orphan tendon cell (G), and a direct muscle attachment in the process of detachment (H); in all cases, junctional surfaces and their intracellular linkage to the cytoskeleton seem fully intact. Bar, 10 μm (A–D), 6.1 μm (E), and 900 nm (F–H).

Figure 4.

Ultrastructural features of late larval direct muscle attachments. (A–C) Image of a direct muscle attachment in sagittal view (A) and two close-ups (B and C; as boxed in A). In this preparation, the tendon cell (asterisk) occurs more translucent than surrounding epidermal cells (ep), making its maximal extension (between black curved arrows) easy to visualize. Arrays of microtubules (between black and white arrowheads; open white arrows point at microtubules) are longer toward the distal tip of the myotendinous junction, and the inner layer of the cuticle is bent upward in this area (double chevron in A). Basement membrane reaches to the distal (white arrows in A and B) and proximal (black arrows in A and C) end of the myotendinous junction, but it seems not to penetrate the myotendinous cleft (open black arrows). (D–E′) The cytoskeletal array of another tendon cell showing a relatively broad band of densely coated finger like indentations apically (black arrowhead in D) and the undulating band of the hemi-adherens junction basally (white arrowhead in D); these electron-dense specializations are likely to be the structural equivalents to the apical and basal bands labeled by actin::GFP (ActGFP) or phalloidin in confocal images of tendon cells (arrowheads in E). Bar, 2 μm (A), 0.8 μm (B and C), 1.5 μm (D), and 7.4 μm (E–E′).

Figure 5.

Properties of F-actin arrays in tendon cells. (A¹–E³) Different views of control (con), trypsin-(Tryp), and/or cytochalasin D-treated (CytD) direct muscle attachment sites; planes of view (bottom right) and symbols as explained in Figure 1B. Specimens are labeled with phalloidin (Phal) together with tendon cell-specific actin::GFP (actGFP). The organization of actin into a cytoskeletal belt (between arrowheads in A³) and dashed/stippled fields (A⁴) clearly resembles tubulin::GFP-labeled specimens (compare Figure 3A). Cytochalasin D destroys most actin in these cells (dots in B² and E²), but it has little or no effect on the F-actin arrays (B³ and B⁴), even in orphan tendon cells (E³). Muscle detachment does not significantly affect F-actin arrays in cytoskeletal belt (C³) and stippled fields (C⁴), and F-actin arrays can still be identified, if detached tendon cells are cultured for another 2 h (D³). (F–F″) Close-up of a detached tendon cell labeled with tubulin::GFP and phalloidin showing the closely intermingled nature of both cytoskeletal fractions. Bar, 10 μm (A–E), 2.8 μm (F′–F″).

Figure 6.

Further properties of tendon cells. Direct muscle attachments (white dots, muscles; asterisks, and tendon cells), labeled as indicated bottom left: Sqh, Spaghetti squash; Zip, Zipper; Ena, Enabled; EnaGFP, Enabled::GFP; act, actin::GFP; Phal, Phalloidin; anterior is left. (A–B′) Sqh::GFP localizes to the cytoskeletal belt (between arrow heads in A¹) and the apical fields of cytoskeletal arrays (A²); Zipper colocalizes with Sqh at the cytoskeletal belt (B). (C) Enabled is enriched in late larval tendon cells (asterisk), but it does not concentrate at cytoskeletal arrays. (D) Enabled::GFP is also absent from cytoskeletal arrays. (E–H) In late embryonic tendon cells, Ena accumulates at the tendon cell periphery of wild type (white arrows in E) but not ena^GC1/23 mutant embryos (F); absence of Enabled has no obvious effect on the F-actin arrays (G vs. H). Bar, 10 μm (A, C, and D) 6 μm (B), and 7 μm (E–H).

Figure 7.

Persistence of F-actin arrays in the absence of Rho1 or integrin function. Live images of embryos (wt, wild type; mys, myospheroid mutant; Rho-IR/RhoN19, tendon cell-specific expression of Rho1 iRNA/RhoN19) with tendon-cell specific expression of GFP-tagged proteins (act, actin; tub, tubulin); anterior up in A–C and left in D–F′. Cytoskeletal arrays are hardly affected by loss of Rho1 function (C and D) and are arranged into rings in myospheroid mutant embryos (arrowheads in D′, E′, and F′). Most pictures are taken of indirect muscle attachments in the dorsal muscle field, where the cytoskeletal arrays form one anterior and one posterior cytoskeletal belt (arrows in D). Bar, 14 μm.

Figure 8.

Persistence of F-actin arrays in the absence of Shot function. Flat dissected embryos (arrows, bottom left, indicate anterior) with tendon–cell-specific expression of actin-GFP (green); muscles are labeled with phalloidin (magenta); different genotypes (indicated on left) were treated for 90 min in either PBS (left column) or PBS with toxin (right column; LaA, 6 μM latrunculin A; CyD, 4 μM cytochalasin D). Arrowheads indicate elongated F-actin arrays. Bar, 19 μm.