Caldesmon Regulates Axon Extension through Interaction with Myosin II*

Background: Axon extension, an essential step for creating neural circuits, is regulated by cytoskeletal dynamics. Results: Caldesmon is a regulator of the actin cytoskeleton and enhances axon extension through direct interaction with myosin II. Conclusion: Caldesmon binding to myosin II inhibits myosin II function, resulting in the enhancement of axon extension. Significance: This study elucidates how caldesmon-regulated actin-myosin system is involved in axon extension. To begin the process of forming neural circuits, new neurons first establish their polarity and extend their axon. Axon extension is guided and regulated by highly coordinated cytoskeletal dynamics. Here we demonstrate that in hippocampal neurons, the actin-binding protein caldesmon accumulates in distal axons, and its N-terminal interaction with myosin II enhances axon extension. In cortical neural progenitor cells, caldesmon knockdown suppresses axon extension and neuronal polarity. These results indicate that caldesmon is an important regulator of axon development.

Neurons in the developing brain extend axonal and dendritic arbors that create a complex circuitry, and the guided extension of axonal fibers is an essential step in this process. Axon extension is regulated by the coordinated interaction of microtubules and actin filaments in the axonal growth cone. A growing body of evidence indicates that microtubule polymerization and stabilization play positive roles in axon extension (1), whereas actin filament roles are more complicated. For example, knocking out Ena/VASP or Cdc42, which positively regulate actin polymerization, causes axonal tract loss (2,3). In contrast, inhibiting the actin nucleation factor Arp2/3 and pharmacologically destabilizing actin filaments enhances axon extension (4,5). Thus, the fundamental details of axon guidance and regulation by actin filaments are not well understood.
Caldesmon (CaD) 2 was first identified as a smooth-muscle protein that binds calmodulin and actin (6). It has since been found to be ubiquitously expressed in smooth muscle and nonmuscle cells, and to regulate Ca 2ϩ -dependent actomyosin contraction (7,8). CaD binds to the side of filamentous actin (F-actin) and inhibits actin-myosin interactions, as revealed by superprecipitation assays and actin-activated myosin ATPase activity (9 -11). CaD binding also stabilizes F-actin filaments by enhancing actin-tropomyosin binding and preventing the actin-severing activity of gelsolin or cofilin (12,13). CaD plays important roles in migration of non-muscle cells via regulating actin-myosin system (8). We recently reported that CaD is involved in detrimental glucocorticoid-induced effects during cortical brain development (14,15): glucocorticoids increase CaD levels, transiently retarding the radial migration of cortical neuronal progenitor cells. We also reported that CaD localizes to neuronal growth cones (16). Thus, it seems that CaD plays multiple important roles in neuronal development. In this report, we demonstrate a novel role for CaD in axon extension via its N-terminal myosin binding sequence.
Cell Culture and Immunostaining-Hippocampal neurons were prepared from rat hippocampi on embryonic day 18.5. The dissociated neurons were plated on poly-L-lysine-coated coverslips, and cultured in glial-conditioned MEM containing 1 mM pyruvate, 0.6% (w/v) D-glucose, and 2% B27 supplement (Invitrogen). The next day, the culture was changed to a neurobasal medium containing 2% B27 supplement and 0.5 mM L-glutamine. Cortical NPCs were prepared from rat cerebral cortex on embryonic day 15.5 (E15.5), cultured as previously described (14), plated on laminin-coated coverslips, and cultured under basic FGF-free conditions to induce their differentiation into polarized neurons. A549 and HEK293T cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Cells cultured on coverslips were fixed using 4% paraformaldehyde and then processed for immunocyto-chemistry. To label F-actin, Alexa 568-phalloidin (Molecular Probes) was added to the secondary antibody solution.
Transfection-Hippocampal neurons prepared from rat embryos on E18.5 were transfected by the calcium phosphate method as described previously (18). In brief, DNA-calcium phosphate precipitates were prepared using a calcium phosphate transfection kit (Invitrogen). The hippocampal neurons were plated on a Nunclon⌬ surface plate (Nalge Nunc Interna-tional) and incubated with the precipitates for 3 h. The transfected neurons were replated on poly-L-lysine-coated coverslips and cultured for 3 to 5 days. A549 and HEK293T cells were transfected using Lipofectamine 2000 or Lipofectamine LTX (Invitrogen).
Expression Plasmids-The coding regions for human l-CaD, its N terminus (1-263 amino acids), C terminus (264 -558 amino acids), and N terminus ⌬21-47 (lacking amino acids 21-47 from the N terminus), and the N-terminal fragments of rat myosin IIA (1-1961 amino acids) and IIB (1-1976 amino acids) were amplified by PCR and subcloned into the highly efficient mammalian expression plasmid pCAGGS. EGFP and Myc tag sequences were fused to the 5Ј-end of the coding sequences. The mcherry-LifeAct expression vector was constructed as previously reported (19).
RNA Interference-Short-interfering RNAs (siRNAs) against rat CaD were transfected into growing cortical NPCs using Lipofectamine RNAi MAX (Invitrogen). MicroRNA (miRNA) plasmids against rat CaD were constructed as previously described (14) and transfected into hippocampal neurons by calcium phosphate precipitation. The targeting sequences and the siRNA and miRNA knockdown efficacy were reported in our previous studies (14,20).
Immunoprecipitation-HEK293T cells with transfected expression vectors were lysed with Triton-X-buffer (0.05% Triton X-100 (pH 7.6), 30 mM Tris-HCl, 50 mM NaCl 5 mM EGTA, 5 mM MgCl 2 , 1 mM ATP, and protease inhibitor mixture for use with mammalian cell and tissue extracts (Nacalai Tesque)). Immunoprecipitation was performed using the earlier-listed antibodies and protein G-Sepharose (GE Healthcare Life Sciences). The Sepharose beads were boiled in SDS-sample buffer to elute the immunocomplexes.

CaD Enhances Axon Extension in Hippocampal Neurons-
CaD, a ubiquitous regulator of the actin cytoskeleton, localizes along actin fibers and in the ruffling membrane (7,8). Here, we found that CaD was located in the soma and growth cones of primary cultured hippocampal neurons, with the strongest expression in the distal axon (Fig. 1, A-D). CaD levels increased for 3 to 7 days in vitro (DIV) (2.3 Ϯ 0.8-fold at 7 DIV versus 2 DIV) while the neurons established polarity and actively extended axons (Fig. 1E). The location and time-course of CaD expression in these cells are consistent with its having a role in axon extension.
We therefore investigated CaD function in neurite outgrowth by overexpressing or knocking down CaD in hippocampal neurons. We used GFP-fused CaD (GFP-CaD), which has the same functions as endogenous CaD (14,20). GFP-CaD dramatically enhanced axon extension but did not significantly affect dendrite length as compared with the control, GFP (Fig.  1, F and G). GFP-CaD also enhanced formation of filopodia-like protrusions from the soma and axon branches (Fig. 1F). These CaD-induced protrusions were composed of concentrated actin filaments and were distinct from the main axonal branches, which were filled with microtubules (Fig. 1H). Knocking down the endogenous CaD decreased axon length, but not dendritic length (Fig. 1, I and J), indicating that CaD accumulates in the distal axon of hippocampal neurons during their development and enhances axon extension.
CaD Regulates Axon Development in Cortical NPCs-To monitor CaD involvement in early events in neurite outgrowth, we used cortical neural progenitor cells (NPCs), which proliferate as non-polarized cells in the presence of basic fibroblast growth factor (FGF) (14,21). Under basic FGF-free conditions, however, NPCs stop proliferating and establish neuronal polarity with MAP2-positive dendrites and a tau1-positive axon ( Fig.  2A). When CaD was knocked down with siRNAs in proliferating NPCs, tau1-staining showed that the establishment of neuronal polarity was significantly suppressed within three culture days under basic FGF-free conditions (Fig. 2, A and B). Even in polarized cells, the length of tau1-positive axons was significantly shortened by CaD knockdown (Fig. 2, A and C), as observed in hippocampal neurons. At an early stage of NPCs differentiation into polarized cells, immature axons were often stained with both anti-MAP2 and anti-tau1 antibodies. In the CaD-knockdown NPCs, some short axons were MAP2/tau1 double positive, suggesting delayed development of these cells. These findings indicate that CaD plays important roles in establishing neuronal polarity and in axon extension in developing NPCs.
CaD-Myosin Interaction Required for Axon Extension-CaD has been reported to bind smooth muscle myosin at its N terminus and F-actin at its C terminus, suggesting that it functions to link these molecules (22). In the growth cone of hippocampal neuronal axons, CaD colocalized with F-actin and myosin IIA/ IIB, the major non-muscle isoforms of myosin II (Fig. 3A). To examine myosin and actin involvement in CaD-induced axon extension, CaD N-and C-terminal fragments (N-CaD and C-CaD) were expressed separately in hippocampal neurons. N-CaD enhanced axon extension like full-length CaD, but C-CaD did not (Fig. 3, B-D), suggesting that CaD interaction with myosin, but not F-actin, is necessary for CaD-induced axon extension. On the other hand, C-CaD, but not N-CaD, induced formation of the filopodia-like protrusions like full- length CaD (Figs. 1F and 3, C and E), suggesting that this effect is dependent on the C-terminal actin binding domains.
Co-immunoprecipitation was used to determine whether non-muscle myosin II, like smooth-and skeletal-muscle myosins, binds to CaD. Because CaD is reported to bind to the S-1 and S-2 regions of smooth and skeletal muscle myosins (23), we examined CaD interactions with myosin IIA or IIB N-terminal fragments, which are composed of a globular head domain, a neck region, and a small tail fragment corresponding to heavy meromyosin (HMM). As with smooth and skeletal muscle myosins, CaD bound to HMM IIA and IIB, and CaD's C-termi-nal F-actin-binding domains were not necessary for these interactions (Fig. 4, A-C).
Previous studies demonstrated that the 27-amino acid sequence in CaD's N terminus (Tyr-21 to Lys-47 in human l-CaD) is necessary for binding to smooth-muscle myosin (24). N-CaD ⌬21-47 fragment, in which this 27-amino acid sequence is deleted, did not interact with HMM IIA, and a CaD fragment including amino acids 1-47 was the minimum required for HMM IIA binding (Fig.  4, C and D). Importantly, N-CaD ⌬21-47 fragment completely lost the ability to enhance axon extension (Fig. 3, C and D), strongly supporting the idea that CaD is accumulated in the growth cone as  an actomyosin component and enhances axon extension through direct interaction with non-muscle myosin II.
N-CaD Exhibits the Same Effect as Blebbistatin-To determine the significance of CaD interaction with myosin, N-CaD or C-CaD was transfected into A549 cells. CaD has been reported to stabilize actin filaments via its C-terminal F-actinbinding domains, causing thick actin fibers to form (12,26). In A549 cells, C-CaD strongly induced thick actin fiber formation (Fig. 5, A and B). On the other hand, cells expressing N-CaD showed significant actin fiber loss and a flat cell shape with prominent lamellipodia (Fig. 5, A and B). These effects were completely lost in A549 cells expressing an N-CaD ⌬21-47 fragment lacking the 27-amino acid myosin-binding sequence (Fig. 5, A and B). Further, these morphological changes were very similar to those found in cells treated with the myosin II-inhibitor blebbistatin (Fig. 5A). These results suggest that CaD binds to myosin at its N terminus, and that it inhibits myosin II function independently of its C-terminal F-actinbinding domains.
CaD Changes Growth Cone Morphology and Myosin II Localization-To determine the function of CaD in growth cones, we observed growth cone morphology and myosin II localization in the hippocampal neurons expressing CaD fragments (Fig. 6). N-CaD inhibited lamellipodia expansion, whereas C-CaD enhanced filopodia formation in growth cones. Full-length CaD induced both lamellipodia retraction and filopodia formation. N-CaD ⌬21-47 had no effect on growth cone morphology.  In GFP-N-CaD-transfected neurons, myosin II staining was slightly diffuse, but distinctly strong in the basal region of the lamellipodia-poor growth cones. In the cells transfected with GFP-CaD and GFP-C-CaD, myosin II was tightly associated with filopodia. N-CaD ⌬21-47 had no effect on myosin II localization. These results indicate that C-terminal actin binding domains enhances actin bundling in growth cones, leading to filopodia formation, with which myosin II associates. On the other hand, N-terminal myosin-binding domain inhibits lamellipodia formation in growth cone, but scarcely have an effect on the myosin II localization.
CaD Enhances Axon Extension by Inhibiting Myosin-To examine how inhibiting myosin function would affect axon extension, hippocampal neurons were incubated with blebbistatin, myosin light chain kinase inhibitor ML-7, or Rho-associated protein kinase inhibitor Y27632, drugs that directly or indirectly inhibit myosin function. All of these drugs, especially the direct inhibitor blebbistatin, significantly increased axonal length compared with the vehicle control (Fig. 7, A and B). In GFP-CaD-transfected hippocampal neurons, however, blebbistatin did not further accelerate axon extension (Fig. 7, C and  D). Coupled with its effects on axon extension, blebbistatin induced morphological changes in the axonal growth cones, inducing a switch from lamellipodial to filopodia-like protrusions (Fig. 7E). In the GFP-CaD-transfected neurons, axonal growth cones displayed a filopodia-like morphology without expanded lamellipodia, and their morphology was not affected by blebbistatin treatment (Fig. 7E). These findings indicate that blebbistatin and CaD enhance axon extension via the same pathway, through which myosin II function is inhibited.

DISCUSSION
CaD is a ubiquitous regulator of the actin cytoskeleton. Most of CaD functional domains that bind F-actin, tropomyosin, and calmodulin are located in its C terminus, and the C-terminal fragment can inhibit myosin ATPase activity and stabilize actin filaments (8, 12, 26 -28). CaD N-terminal region also has a myosin-binding sequence, through which CaD binds to smooth and skeletal muscle heavy meromyosins (23). This binding domain is probably involved in tethering myosin to actin filaments (22,29), but the significance of myosin binding to CaD had been unclear. In our present study, we clearly demonstrated that CaD enhances axon extension through direct interaction with non-muscle myosin II via its N-terminal myosinbinding sequence. N-CaD, which lacks the all C-terminal functional domains, exhibited the same effect on axon extension as full-length CaD (Fig. 3, C and D), indicating that axon extension does not depend on the CaD-mediated physical bridge between myosin and actin.
In addition to axon extension, CaD induced formation of the filopodia-like protrusions from soma and axon branches (Figs. 1F and 3E). CaD also enhanced filopodia formation in growth cones (Fig. 6). C terminus of CaD, but not N terminus, was required for both functions. C terminus contains some actin binding domains, which are necessary for stabilization of actin bundles (7,8), and the filopodia-like protrusion were composed of concentrated actin filaments (Fig. 1H). These indicate that actin stabilization by the C-terminal domains facilitates formation of these filopodial protrusions independently of N-terminal myosin binding domain.
Results of our experiments using myosin II ATPase inhibitor blebbistatin strongly suggest an inhibitory effect of N-CaD on myosin II function in hippocampal neurons and non-neuronal A549 cells (Figs. 5 and 7). However, previous in vitro study showed that the CaD 1-597 fragment, which lacks the C-terminal actin-binding domains, does not inhibit actin-activated myosin ATPase activity (29). Velaz et al. (22) reported that CaD inhibits actin-activated myosin ATPase activity via its C-terminal F-actin-binding domains, by preventing the myosin head from binding to actin in vitro (21). Considering the discrepancy between these in vitro studies and our in vivo study, we propose that N-CaD inhibits myosin II function by unknown mechanisms, which may include interacting with or recruiting additional myosin-inhibitory factors. Further investigations are required to clarify how N-CaD inhibits myosin II function.
Growing evidences indicate that myosin II function is important for axon outgrowth and axon guidance (30 -35). However, the molecular mechanism underlying actomyosin-mediated axon extension has not been fully evaluated. An early study by Letourneau et al. (36) clearly demonstrated that both "push" by microtubules and "pull" by actomyosin in the growth cone play central roles in axon extension. Actually, actin destabilization by cytochalasin D or ADF/cofilin and myosin II inhibition by blebbistatin enhance axon extension (Refs. 5, 25 and the present study). CaD may inhibit the traction force generated by the actomyosin contraction, thereby augmenting the pushing force from microtubule extension.