Cell Chirality is Reversed from CW to ACW Bias during Cell Attachment.
Attachment to cell-adherent substrate is the prerequisite for expression of cell chirality 14,17. We first took time-lapse microscopy to record cell spreading and nuclei rotation during spreading of human foreskin fibroblast (HFF-1) on culture dish. Cells spread from circular shape to its natural shape from 0 min to 240 min (Fig. 1a and Supplementary Movie 1). To assess the cell chirality, cell nuclei were tracked using the binarized nuclei images to record the trajectory of nuclei movement with 5 min interval (Fig. 1c and 1d). After recording, the chirality is quantified as the probability of CW and ACW rotation within every 15 minutes for all cells. Our results reveal that nucleus rotation begins with CW bias after initial attachment but is reversed to ACW-biased with time (Fig. 1e). Noticeably, during the temporal change of cell chirality, the cell projected area increases (Fig. 1f) while the circularity decreases as cells are approaching to its natural shape (Fig. 1g, h), suggesting a possible dependence of CW/ACW bias to them. Because cells with non-circular shape would inhibit expression of chiral rotation of cells,12 the increased projection area is more likely the reason of chirality reversal. To clarify it, the probability of ACW/CW rotation was further analyzed against the averaged projected area. For circularity greater than 0.7, the reversal of chirality was found (Fig. 1i). In contrast, it is not as apparent for cell circularity below 0.7 (Fig. 1j). Together, the results suggest that the chiral nucleus rotation may require circular cell shape and can be revered from CW to ACW with increase of cell projected area.
Cell Chirality is Reversed from CW to ACW with Increment of Micro-island Area.
To provide more direct evidence of the role of cell projected area to the chirality reversal, we applied arrays of fibronectin micro-islands with different circular areas, surrounding by cell-repellent Pluronic coating (Fig. 2a). Time-lapse measurement of nucleus rotation of HFF-1 fibroblasts was again used to record nucleus rotation on micro-islands. Interestingly, when cells spread on 500 µm2 pattern, the nucleus rotation is neutral (Fig. 2b). Starting from 750 µm2, nucleus rotation becomes CW-biased, i.e., 45.83 ± 1.96% ACW with p-value < 0.01, which is consistent with the chiral bias of cell during initiate spreading. From 1000 µm2 to 2500 µm2, the chiral nucleus rotation is changed to neutral and then reversed to ACW-biased, i.e., 63.43 ± 2.90% ACW with p-value < 0.0001 on 2500 µm2 of micro-islands. Thus, the results on micro-islands provide a direct and consistent evidence showing cell projected area as the key factor for chirality reversal. In contrast, for cells on other non-circular shapes, e.g., triangle, square, and rectangle with area of 750 µm2 or 2500 µm2, the reversal of cell chirality from small to large island is not seen, indicating circular islands as a better platform for exhibition of area-dependent chirality reversal (Supplementary Fig. 1). Furthermore, to test the universality of such phenomenon, we have analyzed the nucleus rotation on two other cell types (Fig. 2c-d), C2C12 mouse myoblasts and human mesenchymal stem cells (hMSCs). They all show the same dependence of chirality reversal based on cell spreading area, demonstrating the universality of this relationship.
Actin Distribution upon Drug Treatment Reveals the Role of Actin Polymerization for ACW Chirality.
To investigate the mechanism, we next examined actin cytoskeleton on such islands. We have previously found that actomyosin cytoskeleton may function as a built-in machinery underlying the cell chirality 14,17. At the micropattern boundaries, actin filament even arises as a chiral swirling pattern composed of transverse arc and radial fibers 28,29 when cultured on micropatterned circles 22. By staining actin filaments for HFF-1 fibroblasts on large islands (2500 µm2), actin forms a swirl pattern of radial fibers in conjunction with transverse arc and can be seen as a ring shape after image stacking (Fig. 3a). Using automated image processing to analyze the tilting angle of actin filament with respect to the radial axis of each cell 23, a ACW bias was found (Fig. 3b), which is consistent with previous findings 22 and the nucleus rotation mentioned above. Interestingly, with nucleus removed by treatment of cytochalasin D followed by centrifugation and recovery in the growth medium, the ACW bias of actin filament is maintained, suggesting actin as the main driver of chiral nucleus rotation (Supplementary Fig. 2). Surprisingly, while CW rotation is observed for cells on small islands (750 µm2), only ventral fibers and transverse arc remain, and radial fiber is rarely observed (Fig. 3c). After image stacking, the actin ring is replaced by accumulation of actin at cell center. Together, it suggests that actin is the major driver of cell chirality, but not necessarily to be solely dependent on the presence of radial fibers.
Such finding is unexpected as previous findings suggested that the chiral swirling of actin filament was initiated from a possible axial rotation of radial fibers during polymerization 22. To further investigate it, we applied a series of cytoskeleton-related inhibitors on HFF-1 fibroblasts on two representative micro-islands, 750 µm2 as small island inducing CW bias and 2500 µm2 as a large island inducing ACW bias. We have identified A23187 Calcium Ionophore (A23) that stimulates perinuclear actin formation 30, the inhibitor of F-actin polymerization Latrunculin A (LatA) 31, RAC1 inhibitor NSC23677 (NSC) 32, formin FH2 domain inhibitor SMIFH2 33,34, and activator of actin-based complexes Lysophosphatidic acid (LPA) 35. On large islands (2500 µm2), the original ACW chirality (63.43 ± 2.90% ACW) remains unchanged with A23 treatment (Fig. 3d) and becomes neutralized by LPA treatment and even more ACW-biased with NSC and SMIFH2 (Supplementary Fig. 3a). In contrast, treatment of LatA could reverse the chirality to CW bias (45.58 ± 2.15% ACW, Fig. 3d). On the other hand, the CW bias on small islands (750 µm2) is unaffected by the treatment of LatA (Fig. 3e), but is neutralized by NSC, SMIFH2 and LPA (Supplementary Fig. 3b). More interestingly, A23 reverses the nucleus rotation from its original 45.83 ± 1.96% ACW to 54.00 ± 1.60% ACW bias (Fig. 3e). From the above results, treatment of A23 and LatA appear to be effective to reverse the induced CW/ACW bias on small/large island, respectively.
We next analyzed the actin distribution of HFF-1 fibroblasts in response to the A23 and LatA treatment. Comparing to untreated cells, actin is lessened around the nucleus with A23 treatment, but more ventral stress fibers across cell nucleus can be seen with LatA treatment (Fig. 3f). To statistically illustrate actin distribution upon treatment, we applied image stacking. A23 treatment pushes the actin around the nucleus toward the cell edge (Fig. 3g), while the LatA treatment, to the opposite, brings actin closer to the center, leading to more equalized actin distribution. For cells on small islands, similar effects are seen, and more interestingly, radial fibers are restored with A23 treatment (Fig. 3h, i). To highlight the difference, we used differential heat map by subtracting the stacked actin image between treated and untreated cases. Clearly, A23 concentrates actin at the cell edge but lessen it at the center (Fig. 3j) and a distinct peak of actin intensity appears immediately adjacent to the cell edge (dash line) (Fig. 3k). In contrast, for LatA treatment, actin is concentrated around the nucleus while the peripheral ring of actin is lessened than that of untreated control (Fig. 3j-k). Same effects are also seen in cells on small islands (Fig. 3l-m). Because polymerization of radial actin fibers arises from nucleation of actin filament at their barbed end near the leading edge, such observation suggests an increased polymerization of radial fibers upon A23 treatment which may account for the ACW chiral rotation, especially for cells on smaller islands where their original CW rotation can be reverted to ACW rotation upon restoration of radial fibers. On the other hand, as an inhibitor of F-actin polymerization, LatA appears to suppress the polymerization of radial fibers at cell edge, which may underlie the induced CW nucleus rotation of cells on large islands with LatA treatment.
CW-Swirling of Transverse Arc Acts as the Main Driver for CW Chirality.
How does actin polymerization near the cell edge affect the cell chirality? To elucidate the mechanism, we conducted transfection of LifeAct in HFF-1 fibroblasts. Time-lapse video microscopy shows centripetal-growing radial fibers originate from the cell periphery (Fig. 4a and Supplementary Movie 2, 3). With increase of length, the radial fibers start to tilt rightward accompanying with retrograde flow along the right-tilted growth direction of radial fibers which eventually drives an overall ACW rotation of entire cytoskeleton around cell nucleus. As such, the actin filament forms a chiral swirling pattern with strong coherence of nucleus rotation. Such inward growing filament with rightward turning is consistent with the literature 21,22 and can be explained by the righthanded double helical structures of actin 36. As the barbed ends of actin filament is capped and tethered by formin, incorporation of new actin monomer to the righthanded double helix would cause the other free end of actin filament growing with righthanded axial spinning (Supplementary Movie 4) 22. In contrast, for cells on small islands, the nucleus rotation is solely driven by the retrograde flow of transverse arc without apparent radial fibers (Supplementary Movie 5).
On the other hand, for cells with LatA treatment, while some radial fibers are seen, such rightward turning is absent (Fig. 4b and Supplementary Movie 6). Instead, we noticed a very organized CW swirling of transverse arc that approaches to the nucleus during the retrograde flow, eventually becoming a CW-bias swirling wrapping that causes the nucleus to commit to CW rotation (Fig. 4b and Supplementary Movie 6). Such active transverse arc tends to connect radial fibers such that ventral stress fibers across cell nucleus are formed. As a result, for cells on large island with LatA, more actin is concentrated around the nucleus (Fig. 3j). In some cases, the CW-bias of nucleus rotation occurs with this CW-bias swirling wrapping of transverse arc without distinct, bundled radial fibers (Supplementary Movie 7).
To further prove the role of retrograde flow of transverse arc in CW rotation, we perturbed the major components of transverse arc. Transverse arc are curved actomyosin bundles 28 formed end-to-end annealing of α-actinin and myosin II bundles in a periodic pattern 29. Different from radial fibers, transverse arc are contractile because the incorporation of myosin II bundles which, together with polymerization of radial fibers, drives the retrograde flow of transverse arc 37. Thus, perturbation of α-actinin and myosin II would impose direct influence on the transverse arc. We first applied overexpression of α-actinin, which could promote the bundling of radial fibers and restrict their axial spinning. Meanwhile, it would also promote the assemble of actin filament during transverse arc formation. Our results show an enhanced CW rotation with increasing α-actinin expression (Fig. 4c-d). Next. we applied Y27632, the inhibitor the Rho-associated kinase (ROCK) to down-regulate the myosin light chain phosphorylation 38. Interestingly, myosin II highly enriched in transverse arc in untreated cells is greatly reduced after inhibition by Y27632 (Fig. 4e). More importantly, it enhances the ACW nucleus rotation for cells on both large and small islands, which supports the role of transverse arc in CW bias of nucleus rotation (Fig. 4f). Together, our results suggest the right-turning of centripetal-growing radial fiber drives the ACW rotation, while the retrograde flow of CW swirling of transverse arc is responsible to the CW-biased nucleus rotation.
While the ACW chirality has been explained by the righthanded axial spinning of radial fiber during polymerization 22 (Fig. 5a and Supplementary Movie 4), how such single form of molecular handedness can be unfolded to CW swirling of transverse arc remains intriguingly and unclear. We noticed that myosin II, while presents in the region of transverse arc mostly, does colocalize with radial fibers (Fig. 4e and the zoom-in image). This finding is consistent with previous findings that tropomyosin localizes to radial fibers and coincides with myosin II incorporation 39. Thus, myosin II may be involved in the connection between radial fibers and transverse arc. Interestingly, myosin II and V have been shown as lefthanded spiral motor on the righthanded actin helix 36,40 (Fig. 5b and Supplementary Movie 8). During the actin gliding, the lefthanded translocation of myosin II could, on the other hand, cause lefthanded twirling of an actin filament about its axis 40. Similar phenomenon of lefthanded twirling of actin has been reported in the retraction of filopodia where its right-screw rotation can later lead to righthanded clockwise cell migration 21. Thus, when transverse arc is dominant, the attachment of tethered myosin II, because it simultaneously attaches to fragment of crosslinked arc filaments, could plausibly cause a lefthanded twirling, or axial spinning of radial fibers growing inward (Fig. 5c and Supplementary Movie 9). As a result, opposite to the rightward tilting of radial fibers based on the righthanded spinning during actin polymerization, a short radial actin filament connecting to transverse arc precursors could exhibit a lefthanded spinning and tilted leftward. Eventually, the radial fibers tilting with either rightward or leftward direction would be in conjunction to retrograde flow of transverse arc, therefore causing the overall ACW or CW rotation of nucleus (Fig. 5d, e).
To further investigate the role of myosin II in the formation of CW swirling of transverse, we applied siRNA gene silencing to knockdown myosin binding protein tropomyosin. Tropomyosin has been reported with isoform-specific localization along radial fibers 39. Among the isoforms, Tpm4 on radial fibers is particularly important for recruitment for myosin II as its presence on radial fibers coincides with the incorporation of myosin II. We first tried to observe the distribution of Tpm4 in untreated cells. Tpm4 decorates the radial and ventral actin fibers (Fig. 5f). Interestingly, in the Tpm4 silencing cell, myosin IIa on radial fibers is less condensed (Fig. 5g), indicating the disrupted attachment of myosin II. Consistently, the actin distribution shows a suppression of transverse arc. Noticeably, the transverse arc is more apparent near the periphery of cells but not forming the distinct actin ring as seen in cells of normal condition (Fig. 5h and Fig. 3a). Such result collectively suggests the lost connection of transverse arc precursors to radial fibers such that they fail to move inward through the retrograde flow. More critically, the nucleus rotation of HFF-1 with Tpm4 silencing exhibited much greater ACW chirality both on large patterns (from 63.43 ± 2.90% ACW to 81.64 ± 2.79% ACW) and small (from 45.83 ± 1.96% ACW to 61.37± 3.09% ACW) (Fig. 5i). In contrast, Tpm1, Tpm2 and Tpm3 knockdown did not show noticeable effect on chirality (Supplementary Fig. 5). Based on the above results, we can conclude that the attachment of myosin II on radial fibers via Tpm4 is important for conducing CW swirling of transverse arc.
Endogenous Factor Regulates Chirality Reversal to Differentiation.
We next explore the existence of endogenous factor that could intervene the balance of radial fibers and transverse arc and reverse the cell chirality in HFF-1 fibroblasts. Compared to mDia1 that is mostly responsible for the actin polymerization at barbed end 41, mDia2 promotes the nucleation of Tpm4-decorated actin filaments, which later recruit myosin II before assembling to α-actinin-crosslinked actin filaments 42. Thus, mDia2 may more likely be the upstream factor to tilt the balance between radial fibers and transverse arc. HFF-1 cells with siRNA interfering mDia2 induces an overall reduction of mDia2 expression (Supplementary Fig. 6a). Very importantly, mDia2 silencing reduces the myosin IIa on transverse arc and depletes the co-localization of myosin IIa on radial fibers (Fig. 6a), indicating that mDia2 silencing down-regulates not only the recruitment of myosin II to the assembly of actomyosin bundles but also the connection between radial fibers and transverse arc. With such effect, the cell chirality is tilted to the dominance of radial fibers, which is shown by more ACW bias on both large islands (from 63.43 ± 2.90% ACW to 67.79 ± 3.04% ACW) and small islands (from 45.83 ± 1.96% ACW to 50.88 ± 1.67% ACW) (Fig. 6b). Interestingly, such enhanced ACW bias together with the actin distribution (Supplementary Fig. 6b-c) is similar to the results of formin FH2 domain inhibition by SMIFH2 (Supplementary Fig. 4). To further examine the role of mDia2 in CW rotation, we used transfection of GFP-mDia2 for over expression assay. Interestingly, overexpression of mDia2 causes an overall increase of mDia2 (Supplementary Fig. 6a) and enhanced CW bias of cells on both sizes of islands (Fig. 6c). For the actin distribution, the dominance of transverse arc is seen (Supplementary Fig. 6d-e), and an increase of actin at cell center (Supplementary Fig. 6e) resembling the effect of LatA treatment is observed (Fig. 3g, j). Together, the results suggest that mDia2 as one of the upstream endogenous factors to tilt the balance between radial fibers and transverse arc.
We next explored the potential regulation of cell differentiation by adjusting the cell chirality. Previously we have shown that hMSCs would early committed to a CW chirality with adipogenic induction, which may possibly serve as the mechanical precursor to engage the cell phenotype toward the adipogenic lineage 23. To explore whether the expression of mDia2 can guide the lineage specification, we applied mDia2 silencing to hMSCs (24 hours of transfection and 12 hours of recovery in growth media) before adipogenic and osteogenic dual induction and count the lineage specification by Oil Red O staining (for adipocytes) and Fast Blue staining (for osteoblasts) (Fig. 6d). Interestingly, the transient silencing conducted before induction can still result in an increased ACW bias (from 49.54 ± 4.68% ACW to 56.75 ± 2.82% ACW) after 6 days of dual induction (Fig. 6e), suggesting a long-term effect of mDia2 silencing. More importantly, mDia2 silencing causes a reduced percentage of adipocytes (Fig. 6f), which is consistent with literature that early committed CW bias would induce adipogenesis 43, suggesting the role of cell chirality in guiding the lineage specification.