Annealing 

Figure 4. Contraction of transverse arcs: Contraction of the actin-myosin system allows the filaments to adopt a more regular structure that is contractile along its entire length; the extent of actin filament crosslinking may regulate these contractions. Myosin II motor protein activity is necessary for complete transverse arc formation (figure adapted from [2]).

As the crosslinked filaments are joined together to form long bundles, transverse arcs are fortified with myosin II and the width between alternating bands of αあるふぁ-actinin and myosin thick filaments is equalized [2]. Although not directly demonstrated, it is reasonable to suggest that the actin-myosin interaction and initiation of myosin II-dependent contractions fosters this alignment [37]. In line with the concept that contractile activity influences the structural organization of the cytoskeleton, the transverse arcs disassemble when the actin filaments are no longer held in place by contractile tension and the myosin II bundles [2]. Furthermore, both contraction strength [38, 39, 40] and regulation of myosin II activity [41] correlate with stress fiber formation and they regulate the overall stability and integrity of stress fibers [42, 43, 31, 44]. The extent of contraction also directly correlates with altered transcription of certain genes [38]. Finally, there is a reciprocal relationship between the strength of contractions and changes in cell morphology: cell morphology contributes to the formation of stress fibers and the resulting magnitude of contractile force that can be generated [45], while in reverse, the extent of contractions will influence the cell morphology [38]. Although transverse arcs are not connected directly to focal adhesions [14], the contractile forces produced in transverse arcs can presumably be transmitted to the cell-substrate interface indirectly through the dorsal stress fibers and their associated focal adhesions. Whether this predicted force transmission actually leads to extracellular remodeling, is relatively unexplored.

Dorsal stress fibers in motile cells are formed from actin filament bundles that are initiated and extended from cell-substrate adhesions at the leading edge (aka focal complexes [FXs]) (reviewed in [46]). Similar to graded polarity bundles, retrograde F-aof nectin flow from the leading edge and myosin II activity may be involved in promoting the formation w adhesions [12]; however, new adhesions may also form in a manner that is either force independent or requires relatively low tension [47]. Relative to the migration of the leading edge, the cell-ECM adhesions are static [16]. Nevertheless, stress fiber adhesion components themselves (e.g., αあるふぁ-actinin, vinculin, and talin [48]) are highly plastic and their dynamics are dependent upon mechanotransduction events (reviewed in [46]): as the cell body moves, FX mature into more stable and less dynamic focal adhesions by force-induced structural rearrangements that promote the exchange and addition of new components to the adhesion (reviewed in [49]). In fibroblasts, actin flow appears to cause the formation of "mini-ruffles" and microspike bundles, which may also play a part in forming the initial adhesions found in dorsal or ventral stress fibers [50]. Inhibition of focal adhesions assembly blocks stress fiber formation in general [10].

Initiation is triggered by the activation of adhesion molecules, such as integrin, and the tension-dependent recruitment and activation of the Rho family of GTPases, including Rac1 (reviewed in [51, 52]). Rac1 activity strengthens the nascent adhesion by promoting both the assembly of scaffolding proteins and the recruitment of actin-binding proteins to the adhesion site. Although the initial polymerization of actin filaments from adhesion sites is regulated by all Rho GTPase family members (e.g., Rac1, RhoA, Cdc42) [53], protein recruitment and stress fiber assembly is primarily modulated by signaling cascades involving RhoA [3] (reviewed in [4]).

The short unipolar actin filaments in dorsal stress fibers are polymerized and elongated from focal adhesions via a formin-dependent mechanism (e.g., mDia1/DRF1) [2]. This leaves the barbed ends of the actin filaments directed towards the adhesion (reviewed in [54]). The enzymatic activity of Rac, Rho, and their effectors (e.g., ROCK) is necessary to recruit formins to focal adhesions, where they are thought to play a vital role in creating a stable pool of cortical actin and in maintaining free filament barbed ends [55, 56, 57].

The elasticity of formins at focal adhesions may be tied to their mechanosensing ability, as suggested by increased force-induced actin polymerization at these adhesive sites [58]. Several studies also show that actin filament bundles in filopodia can serve as precursors of dorsal stress fibers or graded polarity bundles [11, 12]. Certain groups that did not observe actin polymerization during stress fiber formation have suggested that stress fibers may form as a result of F-actin stabilization brought about by filament bundling and merging [10].

Stress fiber assembly at focal adhesions is thought to occur by a general mechanism irrespective of the cell type [59]. Once a nucleus of short actin filaments has been initiated from the focal adhesions by formins, elongation occurs in a centripetal direction (towards the cell body) and the filaments are rapidly bound and stabilized by αあるふぁ-actinin [2, 60]. αあるふぁ-actinin not only crosslinks the filaments into small bundles, but also fortifies the bridges between the cytoskeleton and the plasma membrane through its interaction with adhesion receptors (e.g., integrins [60]). Immature actin bundles in spreading cells have the focal adhesion proteins, vinculin and paxillin, not only at their ends but also along the entire length of the actin filaments [47].

Similar to other cortical actin networks that are composed of actin bundles (e.g., microvilli, stereocilia), the elongation of filaments from the base of the adhesion would presumably contribute to the retrograde flow of actin away from the adhesion and promote the condensation of the dorsal fiber into larger bundles as they are moved towards the cell body [2, 21 , 22]. The filament orientation may also change despite the adhesion site remaining stationary relative to the substrate [16]. Dorsal stress fibers can also be connected to transverse arcs in the cell body where the transverse arcs filaments are suggested to continuously supply the dorsal stress fibers with actin filaments [1]. Precisely how this is achieved, is unknown (reviewed in [36]); however, it likely involves actin binding proteins and myosin motor activity.

Interestingly, the assembly of dorsal stress fibers and transverse arcs appears to be connected: transverse arcs encounter dorsal stress fibers as they are transported towards the cell body via cell-wide actin-myosin contractions [2]. The unipolar filaments in dorsal stress fibers are typically non-contractile (at least in the human U2OS bone cell line); however, because their proximal ends are connected to transverse arcs, myosin II is occasionally incorporated into the ends of the DSF, simultaneously displacing αあるふぁ-actinin in this process [2]. The connection to transverse arcs and maturation of elongating dorsal stress fibers (in terms of length) seem to be prerequisites for the incorporation of myosin into dorsal stress fibers [2]. Transverse arcs are also thought to continuously supply dorsal stress fibers with actin filaments as they retract centripetally (reviewed in [1]). Exactly how transverse arcs filaments are fed into dorsal stress fibers is unknown, but based on experiments using purified components [33], mathematical models, and live cells [61, 30], it has been suggested that myosin bundles may recruit the filaments and facilitate polarity sorting [35] (reviewed in [1, 36]). In line with this concept, factors that regulate myosin bundle formation (e.g., myosin light chain kinase [62], Rho-associated kinase [63]) or their binding to stress fiber actin filaments (e.g., tropomyosin [64]) will likely contribute to incorporation of dorsal stress fibers.

Although transverse arcs are not connected directly to focal adhesions, the contractile tension generated by transverse arcs presumably can be transmitted to the cell surface and transferred to the substrate through the dorsal stress fibers; whether this predicted force transmission actually leads to extracellular remodeling, is relatively unexplored.

Figure 10. Association of ventral stress fibers: Pre-existing dorsal stress fibers interact with transverse arcs to demarcate a region of the arc that will align between the two dorsal stress fibers. Crosslinking of the actin filaments by actin binding proteins assistss this interaction. Actin subunits are exchanged throughout the stress fibers and arcs either as a result of polymerization onto the ends of actin filaments, binding to stress fiber myosin, actin treadmilling or interaction with actin-binding proteins in stress fibers (figure adapted from [2]).

Similar to other cortical actin networks that are composed of actin bundles (e.g., microvilli, filopodia, stereocilia), the elongation of actin filaments in the lamellipodium would presumably contribute to the retrograde flow of actin away from the leading edge and towards the cell body, thereby fostering interactions between the dorsal stress fibers and the transverse arcs [2, 21, 22]. Transverse arcs filaments are thought to supply the dorsal stress fibers with filaments of mixed polarity as they are assembled; how this is achieved is relatively unknown, but based on experiments using purified components [33], permeabilized cells [34] and live cells [24, 30], it has been suggested that myosin bundles may recruit the filaments and facilitate polarity sorting [35] (reviewed in [1, 36]). In line with this concept, it seems reasonable to suggest that the myosin filaments may impact the dynamic interaction between the transverse arcs and dorsal stress fibers and foster their association into ventral stress fibers. 

Figure 11. Alignment of ventral stress fibers: The pre-existing dorsal stress fibers interface with the transverse arcs which leaves the arc filaments aligned between the two dorsal stress fibers. Myosin bundles donated by the transverse arc continue to contract, which helps bring the fused structure into complete alignment.

Once the bundled actin filaments in dorsal stress fibers fully interact with the transverse arc filaments, the bundles become aligned and completely 'fuse' to create a cohesive contractile structure [2]. Although it hasn't been demonstrated experimentally, it is reasonable to suggest that the actin-myosin interaction and initiation of myosin II-dependent contractions will help foster ventral stress fiber alignment in a manner similar to the reorientation of lamellipodial and filopodial actin filaments during cell migration [30, 24, 31, 11, 12]. Factors that impact tension-dependent cell adhesion and actin bundle formation are also expected to be key contributors to ventral stress fiber alignment and maintenance. It should be noted that although transverse arcs formation and stability is strictly dependent upon myosin II activity, the formation and stability of ventral stress fiber appears to be somewhat less dependent [2]. Furthermore, ventral stress fibers are fairly stable and their components exhibit the slowest turnover rates relative to dorsal stress fibers and transverse arcs [2].