Abstract
Spectrin is a membrane skeletal protein best known for its structural role in maintaining cell shape and protecting cells from mechanical damage1-3. Here, we report that spectrin dynamically accumulates and dissolves at the fusogenic synapse, where an attacking fusion partner mechanically invades its receiving partner with actin-propelled protrusions to promote cell-cell fusion4-7. Using genetics, cell biology, biophysics and mathematical modeling, we demonstrate that unlike myosin II that responds to dilation deformation, spectrin exhibits a mechanosensitive accumulation in response to shear deformation, which is highly elevated at the fusogenic synapse. The accumulated spectrin forms an uneven network, which functions as a “sieve” to constrict the invasive fingerlike protrusions, thus putting the fusogenic synapse under high mechanical tension to promote cell membrane fusion. Taken together, our study has revealed a previously unrecognized function of spectrin as a dynamic mechanoresponsive protein that shapes the architecture of intercellular invasion. These findings have general implications for understanding spectrin function in other dynamic cellular processes beyond cell-cell fusion.
The mechanical properties of cells are dynamically controlled in many cellular processes, such as cell division, fusion, migration, invasion, and shape change. Spectrin is best known as a membrane skeletal protein critical for maintaining cell shape and providing mechanical support for plasma membrane1-3. In erythrocytes and neurons, spectrins, together with actin, ankyrin and associated proteins, form a static polygonal lattice structure8-10 or an ordered periodic longitudinal array11 underneath the plasma membrane to protect cells from mechanical damage12. Such a mechanoprotective function of spectrin is made possible by holding the spectrin network under constitutive tension13. However, in many cellular processes, mechanical tension is generated upon transient cell-cell interactions. How spectrins, which are expressed in most eukaryotic cells, respond to transient and localized mechanical stimuli in dynamic cellular processes remains largely unknown.
Cell-cell fusion is a dynamic process that occurs in fertilization, immune response, bone resorption, placenta formation, and skeletal muscle development and regeneration14,15. It has been demonstrated in a variety of cell fusion events from Drosophila to mammals that cell fusion is an asymmetric process4-7,16. At the site of fusion, known as the fusogenic synapse, an attacking fusion partner invades its receiving fusion partner with actin-propelled membrane protrusions4-7,16, while the receiving fusion partner mounts a myosin II (MyoII)-mediated mechanosensory response6. The invading and resisting forces from the two fusion partners, in turn, push the two cell membranes to close proximity and put the fusogenic synapse under high mechanical tension to promote fusogen engagement and cell membrane merger5,6. Although previous studies have shown that multiple long (1-2 μm) and thin (~200 nm in diameter) invasive protrusions from the attacking fusion partner are required for cell-cell fusion4,5,17,18, it is unclear how these protrusions are spatially constricted and shaped in order to provide high mechanical tension at the fusogenic synapse.
In a genetic screen for new components required for cell-cell fusion, we have uncovered an essential function for spectrin in Drosophila myoblast fusion. Spectrins are composed of α- and β-subunits, which laterally associate to form antiparallel heterodimers and then assemble head-to-head to form flexible, rod-like heterotetramers with actin-binding domains at the two ends1-3. The Drosophila genome encodes one α-, β- and βheavy(H)-subunit, the latter of which (also known as karst) is homologous to βV-spectrin in vertebrates19. Zygotic mutants of α- or βH-spectrin exhibited minor myoblast fusion defects (Fig. 1a-h, i) likely due to maternal contribution, whereas α/βH-spectrin double mutant showed a severe fusion defect (Fig. 1e, i). The functional specificity of α/βH-spectrin in myoblast fusion was demonstrated by a genetic rescue experiment, in which full-length βH-spectrin expressed in all muscle cells rescued the myoblast fusion defect in βH-spectrin mutant embryos (Fig. 1f, i), and by an overexpression experiment, in which a dominant negative form of βH-spectrin (mini-βH-spectrin, deleting 15 of the 29 spectrin repeats)20 expressed in muscle cells enhanced the fusion defect of βH-spectrin mutant embryos (Fig. 1g, i). Furthermore, expressing βH-spectrin specifically in the receiving fusion partners, the muscle founder cells, rescued the fusion defect (Fig. 1h, i), demonstrating that α/βH-spectrin functions specifically in the receiving cells.
To determine the subcellular localization of α/βH-spectrin, we performed antibody-labeling experiments using anti-α- and βH-spectrin antibodies in wild-type embryos (Fig. 1j, k), and an anti-Flag antibody in a protein trap line, kstMI03134-GFSTF.1, which carries 3xFlag-tagged βH-spectrin (Fig. 1l). Both α- and βH-spectrin are highly enriched at the fusogenic synapse marked by an F-actin-enriched focus, whereas no enrichment was observed for β-spectrin, which was expressed at a high level in epithelial cells (Extended Data Fig. 1a). The βH-spectrin enrichment at the fusogenic synapse largely colocalized with that of Dumbfounded (Duf), an Ig domain-containing founder cell-specific adhesion molecule21 (Fig. 1j-l). Moreover, expressing a V5-tagged βH-spectrin specifically in founder cells led to the accumulation of βH-spectrin-V5 at the fusogenic synapse (Fig. 1m). In contrast, expressing βH-spectrin-V5 specifically in the attacking fusion partners, the fusion competent myoblasts (FCMs), did not result in its accumulation at the fusogenic synapse (Fig. 1n). These results further support the functional requirement for α/βH-spectrin in founder cells.
To investigate whether βH-spectrin forms a stable membrane skeletal network at the fusogenic synapse, we performed live imaging experiments in Drosophila embryos. Surprisingly, instead of forming a stable network, βH-spectrin exhibited dynamic accumulation and dissolution at the fusogenic synapse accompanying the appearance and disappearance of the actin-enriched invasive structure from the FCM, known as a podosome-like structure (PLS)4 (Supplementary Video 1; Extended Data Fig. 2a, b). Previous studies have shown that the density of an actin focus correlates with its invasiveness, such that loosely packed actin filaments are less invasive, as in dpak3 mutant embryos18. We found that the amount of βH-spectrin accumulation correlated with the density of the F-actin foci, as shown in both wild-type (Supplementary Video 1; Extended Data Fig. 2a, b) and dpak3 mutant embryos (Supplementary Video 2; Extended Data Fig. 2c, d). Specifically, βH-spectrin accumulation in dpak3 mutant embryos was overall weaker than that in wild-type embryos, except for a few spectrin “hot spots” associated with the occasionally aggregated small F-actin foci (Extended Data Fig. 2c, d). These findings suggest that spectrin does not form a static network during myoblast fusion; instead, it forms a transient and dynamic structure that rapidly changes its density, shape and morphology in response to changes in the invasiveness of the PLS. The dynamic behavior of βH-spectrin at the fusogenic synapse was confirmed by fluorescent recovery after bleaching (FRAP). When fluorescently labeled βH-spectrin was photobleached at the fusogenic synapse as its accumulation reached equilibrium, the fluorescence rapidly recovered with an average T1/2 of 59 ± 18 sec (n=6) and eventually reached 54 ± 9% (n=6) of the pre-bleaching level (Fig. 2a, b; Supplementary Video 3). These results suggest that more than half of βH-spectrin at the fusogenic synapse is in a mobile fraction and that βH-spectrin dynamically associates and dissociates at the fusogenic synapse, rather than being maintained in a stagnant structure.
Given the correlation between spectrin accumulation and PLS invasiveness, we tested whether βH-spectrin accumulation at the fusogenic synapse is triggered by mechanical force generated by the invasive PLS or is recruited by the cell adhesion molecules in the founder cell, Duf and its functionally redundant paralog Roughest (Rst)21,22. We expressed a truncated form of Duf lacking its entire intracellular domain (DufΔintra) (Fig. 2c) in duf,rst double null mutant embryos and asked if βH-spectrin could still accumulate. Remarkably, βH-spectrin exhibited significant accumulation at the fusogenic synapse marked by the invasive F-actin foci and DufΔintra enrichment, albeit with a slight decrease in intensity compared with the wild-type control (Fig. 2d, f, g). In contrast, Antisocial (Ants)/Rols7, a founder cell adaptor protein that binds to the intracellular domain of Duf23-26, did not accumulate at the fusogenic synapse (Fig. 2, compare e and f; Fig. 2g). Thus, βH-spectrin accumulation in the founder cell can be triggered by invasive force from the PLS, independent of the chemical signaling from cell adhesion molecules. Consistent with these results, when cultured Drosophila S2R+ cells were induced to fuse by expressing the FCM-specific adhesion molecule Sticks and stones (Sns) and the C. elegans fusogen Eff-15, βH-spectrin also accumulated to the fusogenic synapse in response to the invasive protrusions, despite the lack of endogenous Duf and Rst in these cells5 (Fig. 2h).
To test directly whether βH-spectrin exhibits mechanosensitive accumulation, we performed micropipette aspiration (MPA) assays, in which a pulling force is applied to the Drosophila S2 cell cortex by a micropipette with a diameter of ~5 μm. Previously, we have shown that myosin II (MyoII) is a mechanosensory protein that accumulates to the tip of an aspirated S2 cell where the mechanical stress is high6. Interestingly, GFP-βH-spectrin also showed mechanosensitive accumulation, although unlike MyoII, GFP-βH-spectrin accumulated to the base area of the aspirated portion of the cell (Fig. 3a, d). This effect was not due to an increase in membranous materials in this area, since an RFP-tagged PH domain27, which interacts with phospholipids on the plasma membrane, did not accumulate at the base (Fig. 3b, d). Nor did this effect depend on the adaptor protein Ants that did not show mechanosensitive accumulation (Fig. 3b, d). In addition, no accumulation was observed for GFP-βH-spectrin-ΔC (Extended Data Fig. 3a), which deleted a C-terminal fragment containing the tetramerization domain28, or GFP-βH-spectrin-ΔN (Extended Data Fig. 3a), which deleted an N-terminal fragment containing the actin-binding domain28 (Fig. 3b, d; Extended Data Fig. 3b), suggesting that the mechanosensitive accumulation of βH-spectrin requires its tetramerization and actin-binding activities. Notably, βH-spectrin enrichment at the base is time- and force-dependent, which increased linearly over time until reaching its peak level at 80-90 sec after the onset of aspiration (Extended Data Fig. 4a) and increased proportionally to applied pressure (Extended Data Fig. 4b). These results suggest that α/βH spectrin binding to the actin network depends on the number of binding sites at a given time rather than additional cooperative activity of the previously bound tetramers, and that the mechanical force applied to the cortical actin network leads to an increase in the binding sites for the α/βH spectrin heterotetramers.
It is intriguing that α/βH-spectrin and MyoII show distinct domains of mechanosensitive accumulation revealed by MPA. Previous coarse-grained modeling suggested that the tip of an aspirated cell corresponds to an area of maximal actin network dilation (or radial expansion), whereas the base area is under maximal shear deformation (or shape change)29. These analyses suggest that MyoII is a mechanosensory protein for actin network dilation, whereas α/βH-spectrin responds specifically to shear deformation. Consistent with the distinct areas of mechanosensitive accumulation of MyoII and spectrin, βH-spectrin remained accumulated to the base area in cells treated with Y27632, a small molecule that decreases MyoII activity by inhibiting MyoII’s upstream activator ROCK (compare Fig. 3c, top and middle row; Fig. 3d), and MyoII remained accumulated at the tip of βH-spectrin knockdown cells (compare Fig. 3c, top and bottom row; Fig. 3e).
The distinct domains of mechanosensitive accumulation of MyoII and spectrin induced by pulling forces prompted us to ask whether they exhibit a similar response to pushing forces. Course-grained modeling of cells invaded by protrusions with a 5-μm diameter predicted clear separation of dilation vs. shear domains along the invasive protrusion, with maximal dilation at the tip area and maximal shear deformation at the base (Fig. 3f). However, when the invasive protrusions became thinner (approaching a diameter of 400 nm), there was a gradual increase of shear deformation at the tip area, where the dilation deformation remained largely the same (Fig. 3g-j). This model predicted that the mechanosensitive accumulations of spectrin and MyoII induced by thin protrusions may overlap at the tip area, although the maximal shear deformation still resided at the base. To test this directly, we performed atomic force microscopy (AFM) experiments, in which a pushing force was applied to cells expressing GFP-βH-spectrin and RFP-MyoII by a cantilever with a diameter of ~200 nm, closely mimicking the length scale of the invasion protrusions at the fusogenic synapse (Fig. 3k). βH-spectrin and MyoII exhibited rapid and largely overlapping domains of accumulation to the indented area by the cantilever (Fig. 3l; Supplementary Video 4), validating the pattern of mechanosensitive response predicted by the course-grained model. Consistent with this, both βH-spectrin and MyoII accumulated in largely overlapping domains together with Duf at the fusogenic synapse in founder cells of Drosophila embryos (Fig. 1k-m)6.
What is the biological function of spectrin accumulation at the fusogenic synapse? In α/βH-spectrin double mutant embryos, the founder cell-specific adhesion molecule Duf and its interacting protein Ants were both dispersed along the muscle cell contact zone between two fusion partners, instead of forming a tight cluster at the fusogenic synapse as in wild-type embryos (Fig. 4a, b; Extended Data Fig. 5a, b). Time-lapse imaging revealed dynamically dispersing Duf at the fusogenic synapse in the α/βH-spectrin mutant embryos (Supplementary Video 5), compared to the tight Duf cluster associated with dense F-actin foci in wild-type embryos (Fig. 4c; Supplementary Video 6). Occasionally, aggregated Duf clusters were initially observed in these mutant embryos, but in these cases Duf gradually diffused over time, suggesting that spectrin is required for the maintenance, but not the initiation, of the Duf cluster (Fig. 4d; Supplementary Video 5).
Since the cell adhesion molecules from the two fusion partners, Duf and Sns, interact in trans at the fusogenic synapse30, we tested whether Duf dispersal in founder cells of α/βH-spectrin mutant embryos affects Sns distribution in the FCM in a cell non-autonomous manner. Indeed, Sns was also dispersed along the muscle cell contact zone at the fusogenic synapse (Extended Data Fig. 5, compare c and d). Moreover, the actin nucleation promoting factors and their interacting proteins, such as the WASP-interacting protein (known as Solitary or Sltr), which are recruited by Sns to the fusogenic synapse31, were dispersed in these mutant embryos (Extended Data Fig. 5, compare e and f). Consequently, the F-actin structure generated by the FCM appeared diffused (Fig. 4b; Extended Data Fig. 5b, d, f), with an average fluorescent intensity of 61 ± 19 per focus on a 0-255 scale (n=35), compared to 170 ± 15 per focus (n=28) in wild-type embryos. Electron microscopy studies revealed stubby and closely abutting toe-like protrusions in α/βH-spectrin mutant, compared to the long, thin and well separated fingerlike protrusions in wild-type embryos (Fig. 4e, f)4,17,18. Thus, Duf restriction by α/βH-spectrin in founder cells non-autonomously regulates Sns localization and the distribution of polymerized actin filaments along the cell contact zone at the fusogenic synapse in FCMs.
To investigate how spectrin maintains the Duf concentration at the fusogenic synapse, we performed co-immunoprecipitation experiments using Drosophila embryos expressing Flag-βH-spectrin and Duf-GFP specifically in the mesoderm. Antibody against Flag, but not a control antibody, co-precipitated both Duf-GFP and α-spectrin, suggesting that α/βH-spectrin restricts Duf via biochemical interactions (Fig. 4g). In the absence of spectrin-Duf interaction, such as in duf,rst mutant embryos expressing DufΔintra, even though the F-actin foci formed initially due to the trans-interaction between DufΔintra and Sns, the condensed foci eventually dispersed at the fusogenic synapse (Supplementary Video 7), similar to the phenotype of the α/βH-spectrin mutant embryos (Fig. 4a, b; Extended Data Fig. 5b, d, f; Supplementary Video 5). Thus, the intracellular domain of Duf is required for stabilizing the mechanosensitive accumulation of spectrin, likely through spectrin-Duf interactions.
Besides restricting cell adhesion molecules, the abnormal morphology of the invasive protrusions in α/βH-spectrin mutant embryos suggests that spectrin may play an additional role in shaping these intercellular protrusions. Using structured illumination (SIM) microscopy, we observed that the spectrin accumulation triggered by initial invasive protrusions did not form an evenly spaced network like in red blood cells and neuronal axons8-10. Instead, the network contained spectrin-enriched patches interspersed with spectrin-free domains (Fig. 4h-k). Interestingly, actin-propelled membrane protrusions only penetrated through the spectrin-free domains but not spectrin-enriched patches (Fig. 4h-k; Supplementary Video 8). The long and thin protrusions triggered additional mechanosensitive accumulation of spectrin at the tip and along the side (Fig. 4h, i), which may constrict further protrusions. These results suggest that the spectrin network in the founder cell functions as a “cellular sieve” to restrict the diameters of the invasive protrusions from the FCM. The resulting long and thin protrusions put the fusogenic synapse under high local tension and promote plasma membrane fusion4,6.
Taken together, this study has revealed a dynamic mechanosensitive response of spectrin and its cell non-autonomous function in shaping the architecture of invasive protrusions. During cell-cell fusion, the mechanosensitive accumulation of spectrin in the receiving fusion partner sets up a transient spectrin-enriched network at the fusogenic synapse, which, in turn, restricts cell adhesion molecules via biochemical interactions and constricts the invasive protrusions to increase local tension and drive the fusion process (Extended Data Fig. 6). What triggers the mechanosensitive accumulation of spectrin? Previous studies using in vitro reconstituted actin networks have demonstrated that mechanical loading/compression increases the density of the actin network32. We suggest that the invasive protrusions at the fusogenic synapse compress and deform the actin network in the receiving fusion partner, thus generating numerous additional spectrin binding sites and leading to its mechanosensitive accumulation. The α/βH-spectrin bound to the dynamic actin network, in turn, may transiently stabilize the network by unfolding the spectrin repeats, as demonstrated in red blood cells under shear stress 33. What makes spectrin dynamic at the fusogenic synapse? In many previously described cellular contexts where spectrin forms a static membrane skeletal structure, the spectrin-actin interaction is stabilized by accessory proteins, such as adducin34 and protein 4.11,2,35,36. Interestingly, adducin and protein 4.1 were both absent at the fusogenic synapse (Extended Data Fig. 1b, c), which may account for the dynamic behavior of α/βH-spectrin during the fusion process. It is worth noting that the fusion-promoting function of spectrin is not fly-specific; it is conserved in mouse myoblast fusion (Extended Data Fig. 7). Given the widespread expression of spectrin in most eukaryotic cell types, our characterization of spectrin as a dynamic mechanosensory protein in fusogenic cells has broader implications for understanding spectrin functions in many dynamic cellular processes beyond cell-cell fusion.
AUTHOR CONTRIBUTIONS
R.D. initiated the project. R.D., J.K., K.S. and E.C. planned the project, performed experiments in Figures 1, 2, 4 (a-d, g) and Extended Data Figures 2, 3, 5, 7, and discussed the data. J.K. and E.C. collaborated with E.S. and D.R. on the MPA experiments and with S.S. and D.F. on the AFM experiments. D.L carried out the SIM experiments. S.L. carried out the EM experiments. T.L. contributed the coarse-grained models. G.T. contributed a spectrin construct. R.D., J.K., K.S. and E.C. made the figures. J.K and E.C. wrote the paper. All authors commented on the manuscript.
AUTHOR INFORMATION
The authors declare no competing financial interests.
METHODS
Fly genetics
Fly stocks used in this study: α-specrg41 (α-spec mutant; Bloomington Stock Center), kst14.1 (βH-spec mutant)19, Df(3L)1226 (βH-spec deficiency line)19, UAS-mini-βH-Spec37, kstMI03134-GFSTF.1(βH-spec trap line; Bloomington Stock Center), dpak3zyg(del) 18, sltrS1946 31, UAS-DufΔintra-Flag38, sns-GAL439, rP298-GAL424, UAS-GFP-Actin and UAS-Actin-mRFP (Bloomington Stock Center). Transgenic flies carrying UAS-V5-βH-Spec, UAS-mCherry-βH-Spec, UAS-Duf-GFP and UAS-Duf-mCherry were generated by P-element-mediated germline transformation. To express genes in muscle cells, females carrying the transgene under the control of an UAS promoter were crossed with twi-GAL4 (all muscle cells), rP298-GAL4 (founder cells) and sns-GAL4 (FCMs) males, respectively. α/βH-spectrin double mutant, α-specrg41,kst14.1/TM6 (labeled as α-spec,βH-spec in figures), was generated using standard genetic methods.
Molecular biology
Full-length βH-spec was amplified by PCR (with or without a tag) from cDNAs generated from mRNA of stage 11-15 w1118 flies. Due to the large size of the βH-spec gene, three fragments were individually amplified using the primers as follows:
βH-spec-5’: GACCGGTCAACATGACCCAGCGGGACGGCATC
βH-spec-3721-3’: CTCCACGAATTCGGTGTCATG
βH-spec-3721-5’: CATGACACCGAATTCGTGGAG
βH-spec-8214-3’: CTCACCCTCTAGAATGCTATTG
βH-spec-8214-5’: CAATAGCATTCTAGAGGGTGAG
βH-spec-3’: CCAAGCGGCCGCTCACTGTGGCGGGACTTGACTC
The three PCR fragments were then subcloned into a Drosophila transformation vector pUAST. To generate the UAS-βH-SpecΔN and UAS-βH-SpecΔC constructs, the following primers were used:
βH-spec-3865-5’: GGAATTCCAACATGGTGTGTCGATCTGCAAATGTTC
βH-spec-8028-3’: GGTCTAGATCACAGCTGATGGGCCTCAGTTAG
To generate pDEST-βH-Spec constructs for GST-fusion proteins for the F-actin co-sedimentation assays, the Gateway cloning system was used (Invitrogen)with the following primers:
βH-spec-1-C:
GGGGACAAGTTTGTACAAAAAAGCAGGCTTCATGACCCAGCGGGACGGCATC
βH-spec-1-K:
GGGGACCACTTTGTACAAGAAAGCTGGGTTTTACTTCTTGCGATCTGCGTCCAT
βH-spec-29-C:
GGGGACAAGTTTGTACAAAAAAGCAGGCTTCGGAGCCAAACAAGTCCACGTC
βH-spec-31-K:
GGGGACCACTTTGTACAAGAAAGCTGGGTTTTATTGGGACGCCGCATTCTGGCG
βH-spec-34-C:
GGGGACAAGTTTGTACAAAAAAGCAGGCTTCCCGAACATGCAACTGCTTAGC
βH-spec-34-K:
GGGGACCAC TTTGTACAAGAAAGCTGGGTTTTATCACTGTGGCGGGACTTGACT dsRNAs were synthesized by in vitro transcription with gene-specific primers containing the T7 promoter sequence (TTAATACGACTCACTATAGGGAGA) at the 5’ end (MEGAscript; Ambion). Synthesized dsRNAs were purified using NucAway™ Spin Columns (Ambion).
Immunofluorescent staining and imaging
Fly embryos were fixed and stained as described previously4,31. The following primary antibodies were used: rabbit α-muscle myosin heavy chain (1:1000) (gift from B. Paterson), rabbit anti-βH-Spectrin (1:100)19, rabbit anti-β-Spectrin (1:400)40, mouse anti-α-Spectrin (1:1; DSHB), guinea pig α-Duf (1:500)4, guinea pig α-Ants (1:1000)23, rat α-Sltr (1:30)31, rat α-Sns (1:500)41, mouse anti-adducin (1:400; DSHB), mouse anti-protein 4.1 (1:400; DSHB), mouse α-Flag (1:200; Sigma) and mouse α-V5 (1:200; Invitrogen). The following secondary antibodies were used at 1:200: Alexa488- (Invitrogen), Cy3-, and Cy5- (Jackson Laboratories) conjugated and biotinylated (Vector Laboratories) antibodies made in goats. For phalloidin staining, FITC- or Alexa568-conjugated phalloidin (Invitrogen) were used at 1:200. Fluorescent images were obtained on an LSM 700 Meta confocal microscope (Zeiss), acquired with LSM Image Browser software (Zeiss) and Zen software (Zeiss), and processed using Adobe Photoshop CS. For quantification of fluorescent signals, the signal intensities of cellular area of interest and control area were measured and processed for presentation by the Image J program (http://imagej.nih.gov/ij/).
Drosophila cell culture
S2 and S2R+ cells were cultured in Schneider’s medium (Gibco) supplemented with 10% fetal bovine serum (FBS) (Gibco) and penicillin/streptomycin (Sigma). Cells were transfected using Effectene (Qiagen) per the manufacturer’s instructions. For RNAi knockdown, cells were first incubated with 3-5 μg/ml of dsRNA for 2-4 days and then transfected with appropriate DNA constructs.
For immunofluorescent staining, cells were fixed with 4% formaldehyde in PBS, washed in PBST (PBS with 0.1% Triton X-100) and PBSBT (PBST with 0.2% BSA) consecutively, and stained with the following antibodies in PBSBT: mouse α-V5 (1:2000; Invitrogen) and rabbit α-GFP (1:1000; Invitrogen). Secondary FITC-, Cy3-, or Cy5-conjugated antibodies were used at 1:400 (Jackson Immunoresearch). To visualize F-actin, FITC- or Alexa 568-conjugated phalloidin (Invitrogen) was used at 1:500 in PBST.
Mouse C2C12 myoblast culture
A pair of predesigned siRNAs against the mouse Sptbn5 gene and negative control siRNA were obtained from Qiagen. RNAi was performed per manufacturer’s instructions. Briefly, approximately 3x105 cells were seeded on each well of a 6-well tissue culture dish and incubated overnight. On day 2, cells were transfected with the mix of two siRNAs against Sptbn5 (200 pmol each) using HiPerFect transfection reagent (Qiagen). On day 3, the cells were washed and differentiated, and cells that were treated in parallel were subjected to RT-PCR to access knock-down (KD) level. Fusion index was determined when the KD level was more than 80% as follows. Three days post-differentiation, cells were fixed and stained with anti-MHC antibodies (MF20) to identify differentiated cells and with CellMaskTM Orange membrane stain (Molecular Probes, Invitrogen) to visualize the overall cell/syncytia morphology. Cells were mounted using Prolong Gold antifade reagent with DAPI (Molecular Probes, Invitrogen) to visualize the nuclei. Ten random 40X microscopic fields under Axioscope 2 (Zeiss) were counted for each experiment and repeated in at least three independent experiments (total number of fields ≥30). The fusion index was calculated as a mean percentage of nuclei in syncytia versus total number of nuclei per field in differentiated cells.
Time lapse imaging and fluorescent recovery after photobleaching (FRAP)
Time lapse imaging of embryos was performed as previously described4. Briefly, embryos expressing fluorescently tagged proteins in muscle cells were collected and dechorionated in 50% bleach. Subsequently, embryos were washed in water, placed onto a double-sided tape (3M), and covered with a layer of Halocarbon oil 700/27 (2:1; Sigma). Time lapse image acquisition was carried out on an LSM 700 Meta confocal microscope (Zeiss).
The FRAP experiments were performed as previously described17. A region of interest (ROI) was manually selected and imaged in 3 to 5 frames to record the original fluorescent intensity (pre-bleach). Then the ROI was quickly bleached to a level at lower than 20% of its original fluorescent intensity and subsequently imaged with an appropriate time interval (post-bleach). The fluorescent intensities of the pre- and post-bleached ROI were measured using the image J program. The Prism software was used to determine the maximum recovery level (the percentage recovery compared to the pre-bleach level) and the half-time of recovery (Thalf) using a kinetic curve fit with an exponential decay equation.
Structured illumination microscopy (SIM)
Stage 13-14 embryos were fixed and stained as described above. The samples were then mounted in Prolong Gold (Molecular Probes) and imaged with an inverted microscope (Ti-E; Nikon) equipped with a 100x oil NA1.49 CFI SR Apochromat TIRF objective lens and an ORCAFlash 4.0 sCMOS camera (Hamamatsu Photonics K.K.). The images were acquired and processed with NIS-Elements AR software (Nikon).
Electron microscopy
Embryos were fixed by the high-pressure freezing and freeze substitution (HPF/FS) method as previously described4,42. Briefly, a Bal-Tec device was used to freeze stage 12-14 embryos. Freeze-substitution was performed with 1% osmium tetroxide, 0.1% uranyl acetate in 98% acetone and 2% methanol on dry ice. Fixed embryos were embedded in Epon (Sigma-Aldrich) and cut into thin sections with an ultramicrotome (Ultracut R; Leica). The sections were mounted on copper grids and post-stained with 2% uranyl acetate for 10 min and Sato’s lead solution43 for 1 min to improve image contrast. Images were acquired on a transmission electron microscope (CM120; Philips).
Recombinant protein purification and F-actin co-sedimentation assay
To purify GST-fused βH-Spec fragments from BL21-DE3 cells (NEB), protein expression was induced with 0.2 mM IPTG at room temperature for 12-15 hours. Cells were harvested and lysed by sonication in the lysis buffer: PBS (pH 7.4), 1% Triton X-100, 5 mM DTT, 1 mM PMSF, and complete-mini protease inhibitor cocktail (Roche). After centrifugation, the supernatant was collected and incubated with pre-equilibrated glutathione-agarose resin at 4°C for 2-3 hours. After washing in the lysis buffer, βH-Spec proteins were eluted with the Elution buffer: 50 mM Tris (pH 7.5), 150 mM NaCl, 5 mM DTT, 10 mM glutathione (Sigma).
F-actin co-sedimentation assay was performed following the manufacturer’s protocol (Cytoskeleton). Briefly, 0.5-1 μM purified proteins were incubated with 4 μM F-actin assembled from monomeric actin for 1 hour in F buffer: 5 mM Tris-HCl (pH 8.0), 0.2 mM CaCl2, 50 mM KCl, 2 mM MgCl2, 1 mM ATP. The F-actin/protein mixtures were centrifugated at 140,000 g for 30 min, and supernatants and pellets were separated and analyzed by SDS-PAGE and Coomassie Blue staining.
Co-immunoprecipitation (Co-IP)
Embryos expressing GFP-tagged Duf and Flag-tagged βH-Spec (kstMI03134-GFSTF.1/twi-GAL4; UAS-Duf-GFP/+) were collected and dechorionated in 50% bleach. After being frozen in liquid nitrogen, embryos were disrupted in cold extraction buffer (50 mM Tris-HCl (pH 8.5), 150 mM NaCl, 0.5% sodium deoxycholate) with 20 strokes in a Dounce homogenizer. After centrifugation at 16,000 g for 20 min, supernatants were removed and incubated with either anti-V5 (control) or anti-Flag (βH-Spec) antibodies at 4°C for 2-3 hours. Protein-G-Sepharose beads (Roche) were used to precipitate antibodies. Precipitated proteins were analyzed by SDS-PAGE and western blotting.
Micropipette aspiration (MPA)
Micropipette aspiration was performed as previously described44. Briefly, a pressure difference was generated by adjusting the height of a motor-driven water manometer. A fixed pressure of 0.4 nN/μm2 was applied instantly to the cell cortex of S2 cells with a polished glass pipette approximately 2-2.5 μm in radius. Images were collected in Schneider’s medium supplemented with 10% FBS on an Olympus IX81 microscope with a 40x (1.3 NA) objective and a 1.6x optivar, utilizing MetaMorph software and analyzed using ImageJ. After background correction, the fluorescence intensity at the sites of protein accumulation was normalized against the opposite cortex of the cell (Ib/Io or It/Io, where Ib, It, and Io is the intensity in the cortex at the base, tip, or opposite side of the cell being aspirated). An ANOVA with Fisher’s least significant difference was applied to determine statistical significance.
Atomic force microscopy (AFM)
S2R+ cells were plated on glass coverslips coated with concanavalin A (Sigma, St Louis, MO) and transfected to express fluorescently tagged MyoII and βH-Spec using Effectene (Qiagen), per the manufacturer’s instructions. Lateral indentation experiments were conducted 2 days after transfection with a modified Catalyst AFM integrated with an Axio Observer fluorescence microscope (Zeiss). To determine the effect of a localized mechanical force on MyoII and βH-Spec localization, the cantilever (MLCT with a pyramidal tip, Bruker) was first brought into full contact, at around 50 nN setpoint force, with the glass surface on a cell-free area within 10 μm from a target cell. Next, the cell was laterally translated into the stationary cantilever using the piezoelectric XY stage and the NanoScope software (Bruker). The cantilever tip indented the edge of the cell by 2–5 μm. Cells were simultaneously imaged with a plan-apochromat 63x/1.4 NA oil immersion objective (Zeiss). Time lapse images were taken at 5 second intervals using the Micro-Manager software (http://micro-manager.org/wiki/Micro-Manager).
Coarse-grained molecular mechanics modeling
In the coarse-grained model, the membrane-cortex composite is represented by a triangulated network where the nodes denote the crosslinking positions and the triangles resemble the meshes in the actin network, which is a network structure composed of proteins such as actin, actin crosslinkers, and MyoII. The system energy of the composite at the coarse-grained molecular level is calculated by where Ebending is bending energy from the plasma membrane; Ein-plane is the in-plane elastic energy associated with the deformation of actin crosslinkers and the dilation of each mesh; Esurface is the surface energy of the whole cell; Evolume is the energy associated with the volume conservation of the cell29,45,46. The motion of the nodes is calculated by the Brownian dynamics equation (or overdamped Langevin dynamics equation) where the driving force is the derivative of the system energy with respect to the local position.
The initial configuration of the system was thermally annealed at room temperature until the fluctuation of the system energy was negligible. This configuration was then mapped to the shape of a receiving cell. For simplicity, the protrusion has a cylindrical shape with the radius r0 and length l on the surface of a cell with a diameter of 10 μm. The tip of the protrusion was a spherical cap of a radius of r0. The final state of the system was achieved after 20 sec of Brownian dynamics simulation with a time-step of 10-5 sec. The area dilation of each node was determined by averaging of the dilation of the triangles with which the node of interest was associated. The shear deformation of each node was calculated in a similar way. The contour of the deformation on the protrusion was plotted by the software Tecplot. The deformations along the length direction of the protrusion were obtained by the extraction tool of the software.
ACKNOWLEDGEMENTS
We thank the Bloomington stock center for fly stocks, B. Paterson for the MHC antibody, Guofeng Zhang for help with HPF/FS, J. Nathans for sharing confocal microscopes, and D. Pan for critically reading the manuscript. This work was supported by NIH grants (R01 AR053173 and R01 GM098816), American Heart Association Established Investigator Award and HHMI Faculty Scholar Award to E.H.C.; NIH grants (R01 GM66817 and R01 GM109863) to D.N.R; NIH grants (R01 GM074751 and R01 GM114671) to D.A.F; NSF grant #MCB-1122013 to G.T.; and NSFC grant #11572316 to T.L. R.D. was supported by an AHA postdoctoral fellowship; K.S. by an AHA Scientist Development Grant; S.S. by a Life Sciences Research Foundation postdoctoral fellowship; and D. L. by a Canadian Institute of Health Research postdoctoral fellowship.