Materials and methods
Cell culture
HeLa cells purchased from ATCC (ATCC CCL-2) were maintained at 37 °C with 5% CO2, in MEM, supplemented with 10% FBS, 1% Penicillin–Streptomycin, and 1% Glutamax (all reagents from Gibco, Canada). For live imaging, cells were seeded on glass-bottom dishes (1.5 thickness; MatTek, USA) and imaged in phenol red-free imaging media.
PM labelling and targeted laser ablation of the PM
CellMask Deep Red Plasma Membrane Stain (Invitrogen, USA) was used to label the PM as per the manufacturer’s instructions and was used to target multiple sequential laser pulses of a MicroPoint UV laser (Andor, Northern Ireland) to a ~ 95 µm2 (11 µm diameter) area of the PM on the side contacting the dish (i.e. basal side). The success of PM ablation was made on the basis of the creation of a visually detectable gap in the fluorescent signal for both the PM (CellMask) and cytoskeletal compartments. Formation of bona fide PM and cytoskeletal wounds (vs photobleaching) was confirmed via a series of independent wounding/repair. FM1-43 (N-(3-Triethylammoniumpropyl)-4-(4-(Dibutylamino) Styryl) Pyridinium Dibromide; Invitrogen, USA) wounding assays [13] (Additional file 1), and re-assessed at regular intervals.
Transfection of genetically-encoded probes for live imaging
Cytoskeletal dynamics and RhoGTPase activity were tracked using a series of genetically encoded probes: F-actin (calponin homology domain of utrophin; PCS2-UtrCH-GFP; Addgene #26737 or PCS2-UtrCH-mCherry; Addgene #26740), myosin-II (RFP-tagged myosin regulatory light chain 1; PCS2-MRLC-RFP), microtubules (GFP-labelled microtubule binding-domain of ensconsin; PCS2-EMTB-3xGFP; Addgene # 26741), active RhoA (RhoA-binding domain of rhotekin; PCS2- γGBD-GFP; Addgene #26732), and Cdc42 (Cdc42 binding domain of N-WASP; PCS2- ωGBD-GFP Addgene # 26734). All plasmids were transfected using PolyJet In Vitro DNA transfection reagent (SignaGen, USA), as per the manufacturer’s instructions.
Imaging
Imaging was performed at the Cell Imaging and Analysis Network (CIAN) imaging facility at McGill University, using a Quorum WaveFX-X1 spinning disk confocal system on a Leica DMI6000B inverted microscope, two Hamamatsu “ImagEM X2 C9100-23B” EM-CCD cameras, and an 63×/1.40 oil objective (lBL, HCX PL APO; 506192). Cells were kept at 37 °C with 5% CO2 using a Live Cell Instruments Chamlide TC environmental control system. Acquisition was performed using MetaMorph Microscopy Automation and Image Analysis Software (Molecular Devices LLC, USA) and consisted of Z-stacks (11 slices, 0.2 µm apart), taken every 10 s for at least 15 min. 1 × 1 binning, 1 gain, and 100 EM gain was used throughout.
Images were further processed using ImageJ software (National Institutes of Health, USA). Quantitation of fluorescence at the wound site, was performed on maximum intensity projections (3 slices, 0.2 µm apart). Average fluorescence intensity was measured at every 10-s time point, at four 95 μm2 sites: the wound site region (FIW) and three control cytoplasmic regions within the cell (FIC). The relative fluorescence intensity at the wound site (FIR) was calculated by dividing FIW by the average of the three FIC. To enable cell-to-cell comparisons, the resulting FIR were then further normalized to the average FIR on the same unwounded cell. Micrographs, kymographs, and orthogonal views were prepared from maximum intensity projections (MIPs) of the entire imaged volume (10 slices, 0.2 µm apart).
Statistics
Statistical analyses were performed using GraphPad Prism 7.04 (GraphPad Software Inc., USA) and consisted of repeated measure 2-way ANOVAs, followed by Dunnett’s multiple comparisons tests. Significance was set at p < 0.005.
Results and discussion
We provide here substantial evidence that under normal 2D-culture conditions, mammalian cytoskeletal repair or large PM and cytoskeletal injuries involves actin polymerisation, without the formation of actomyosin contractile arrays.
Wound healing response involves uniform actin polymerisation, but not myosin-II accumulation
We first set out to also track the spatio-temporal dynamics of actin polymerisation that follows laser-mediated injury of HeLa cells. As expected, the ablated area showed clear indications of actin polymerisation following wounding (Figs. 1 and 3a; Additional file 2, Additional file 3, Additional file 4, Additional file 5, Additional file 6, Additional file 7, Additional file 8 and Additional file 9). Furthermore, UtrCH FIR rapidly rose above the baseline post-wounded levels (significant after 30 s), which was significantly higher than the pre-wounding baseline levels after 6 min (Fig. 1b), suggesting enrichment of F-actin at the wound site. CR-mediated repair involves the formation of two functionally and structurally distinct zones of actin polymerisation: a dense ring of F-actin and Myosin-II located at the wound edge, which is surrounded by a wider array of perpendicularly oriented F-actin [5, 11, 14, 15]. Neither F-actin patterns were readily observable (Fig. 1a, 3a) during the course of more than 37 individual wounding assays. Kymographic and line scan analyses (Fig. 1c, 3b, c; Additional file 2) confirmed this qualitative assessment and also failed to reveal any other significant spatially defined enrichment of the UtrCH signal. Instead, UtrCH signal increased uniformly over the entire area of the wound site, with no noticeable enrichment at the wound edge (Fig. 1c, 3c; Additional file 2). Orthogonal views of the wound site were similarly unable to detect any localized UtrCH signal indicative of a clearly defined actomyosin structure (Fig. 1d).
The spatio-temporal dynamics of myosin-II following injury was also investigated (Fig. 2a), as it is responsible for the actomyosin array’s contractility [11]. We found that the return of myosin-II at the wound site did not follow the same pattern as we observed for F-actin, as the rate of MRLC FIRW signal accumulation to the wound site (Fig. 2b) was much slower than the one observed for UtrCH, and never went above pre-wounded baseline levels within the time allotted (Fig. 2b). Like F-actin, myosin-II did not show sign of accumulating according to any discernible patterns (Fig. 2a, c and d; Additional file 10). The sequence timing of F-actin and of myosin-II accumulation, broadly confirms the findings of other reports for wounds created on both the apical or basal side of cells, using a variety of wounding methods [16,17,18]. Specific differences in the actin polymerisation dynamics between our and all of the other previous studies could be attributable to the differences in wound size, wounding methods (laser vs mechanical), location (apical vs basal), cell shape and in local tension between the different wounding assays.
The mammalian wound healing response does not involve spatially regulated RhoA or Cdc42 signalling, microtubule arrays, or coordination of PM restoration and cytoskeletal repair
The presence of concentric rings of RhoA and Cdc42 activity around the wound edge is a hallmark event of the formation of actomyosin contractile arrays [19]. As such, we set out to characterize the spatio-temporal distribution of RhoA and Cdc42 signalling in the context of our wound healing assay. Neither γGBD (Fig. 3) or ωGBD (Additional file 3) signals coalesced into the clearly defined RhoA or Cdc42 “activity zones” around the wound edge that would be necessary to the formation and contraction of CR, as demonstrated in wounded Xenopus oocytes [19, 20]. Instead, RhoA activity exhibited a pattern that closely matched that of UtrCH and accumulated over the entire area of the wound site (Fig. 3a, c). γGBD signal quickly rose after wounding, but never rose above its “pre-wounded” baseline levels (Fig. 3b). We were unable to observe any quantifiable or observable change in ωGBD following wounding anywhere in the cell, following the wounding event (Additional file 3; Additional file 7). These findings are consistent with a previous study that examined RhoA activity in mouse myoblasts wounded with a single laser blast to the basal side of their PM, which found local RhoA activation signal at the wound site that also failed to show ring-like patterning [18]. The same study attributed this pattern of RhoA activity to the increased reactive oxygen species (ROS) production caused by the increased mitochondrial uptake of calcium through the mitochondrial calcium uniporter (MCU) [18]. Whether mitochondria mediated ROS production also plays a role in the RhoA activation and actin polymerisation events observed in wounds as large as the ones produced within the context of our study remains to be confirmed.
We also failed to observe the formation of a MT array following injury. Instead, MTs seemed to polymerize from the wound-edge inward (Additional file 6; Additional file 11). This would be consistent with the absence of an actomyosin ring as CR formation is further driven by cortical flow and feedback with microtubules (MTs), which accumulate in a radial array around the wound site and the CR in the Xenopus oocyte model [5, 11, 12]. These observations are also compatible with a study of mouse myoblasts wounded with a single laser point, where MTs were also found to be neither enriched nor necessary for repair [18] and with observations about the dynamics of MTs of mechanically wounded fibroblasts [21].
In echinoderm oocytes, Xenopus oocytes, and early Drosophila embryos, PM and cytoskeletal repair are deeply interconnected (reviewed [6]). The contraction of the actomyosin array not only provides a platform for F-actin polymerisation, but also regenerates normal PM composition as it drags PM inwards through the contracting actomyosin array [14]. We did not observe such coordination in the healing rates of the PM and F-actin wounds (Additional file 12), which again suggests that cells from our wounding assay did not heal according to the classical model of actomyosin array-mediated contraction. It should be noted however, that our assays prioritized the visualization of cytoskeletal events, and as such did not set out to verify the presence of a de facto membrane patch as observable in the Xenopus model or of any of the exocytosis or “explocytosis” events that have been shown to be required for its formation [3].
Conclusions
In mammalian somatic cells, repair of cytoskeletal wounds up to 10 μm is not strictly dependent on the assembly and contraction of an actomyosin array.