PerspectiveBiochemistry

Dynein-Independent Functions of DYNLL1/LC8: Redox State Sensing and Transcriptional Control

See allHide authors and affiliations

Science Signaling  25 Nov 2008:
Vol. 1, Issue 47, pp. pe51
DOI: 10.1126/scisignal.147pe51

Abstract

The highly conserved DYNLL/LC8 proteins promote dimerization of a broad range of targets and are essential for the integrity, activity, or both, of many subcellular systems, such as dyneins, myosin V, and apoptotic factors. Defects in DYNLL/LC8 function lead to severe cellular and developmental phenotypes in multicellular organisms, whereas loss-of-function alleles are lethal. DYNLL/LC8 dimer formation may be controlled by various signaling inputs (including pH changes and phosphorylation), and dimerization has been linked to alterations in the enzymatic activity of neuronal nitric oxide synthase and apoptotic control. A recent report now proposes that DYNLL/LC8-driven interactions are also regulated by changes in cellular redox state, which lead to intermonomer disulfide bond formation and ultimately activation of the transcription factor NF-κB.

Dyneins contain multiple light chain subunits that play additional functions in the cell beyond their role in microtubule motor complexes (1). The most prominent example is the conserved 89-residue protein that has variously been termed LC8, DLC1, and DLC8, among others; the currently agreed designations of the two mammalian isoforms are DYNLL1 and DYNLL2 (2). Here, I use the term DYNLL/LC8 when describing studies that do not differentiate the two mammalian isoforms or that focus on nonmammalian orthologs. This protein class was initially identified in axonemal dyneins (35) and is essential for assembly of dyneins that contain two or more motor subunits. In these enzymes, DYNLL/LC8 associates with the intermediate chains that are required for structural integrity of the dynein particle (6), cargo attachment (7), and, in the case of the cytoplasmic dynein, the regulated association with the adaptor and processivity complex dynactin (8). Furthermore, DYNLL/LC8 deficiencies lead to the complete failure of ciliary and flagellar assembly because of defects in dynein-driven intraflagellar transport (9). However, DYNLL/LC8 proteins also function in a broad array of other multimeric protein complexes, such as myosin V (10), neuronal nitric oxide synthase (nNOS) (11), Bim and Bmf apoptotic factors (12, 13), and a complex required for class I transcription in trypanosomes (14). DYNLL/LC8 proteins are even coopted during viral pathogenesis, for example, by rabies virus P protein (15). In addition to a structural role in maintaining protein complex integrity, DYNLL/LC8 may function in mediating the cellular response to various signaling inputs, including activation of nuclear factor κB (NF-κB) in response to alterations in cellular redox state (16). Because of these diverse functions, in multicellular organisms mutations that lead to partial loss of DYNLL/LC8 result in a broad array of pleiotropic phenotypes, including morphogenetic defects and abnormal neurogenesis, whereas null mutations are lethal, not least because they result in excessive apoptosis (17).

The two mammalian DYNLL/LC8 isoforms differ at only six residues and are expressed at very different levels in various tissues. Importantly, these isoforms were not differentiated in early work on this family. Subsequent studies suggested that these proteins show isoform-dependent associations with both actin- and microtubule-based molecular motors and with apoptotic factors (13). However, the concept of isoform-specific functions received a setback when a proteomics analysis revealed that DYNLL1 and DYNLL2 both associate with brain cytoplasmic dynein (18); it had previously been proposed that one isoform bound to myosin V and the other isoform bound to dynein. The functional significance of DYNLL/LC8 isoform diversity may actually be even more complex because a closely related protein (currently termed DNAL4/LC10) that was initially found in axonemal outer arm dyneins exhibits a broadly distributed expression profile in mammals, which is inconsistent with a unique role in motile cilia or flagella (19).

DYNLL/LC8 proteins function as dimers, with each monomer consisting of an N-terminal β strand followed by two helices and four additional strands (20) (Fig. 1). The dimer interface is formed from two strand-switched antiparallel β sheets, and dimer formation yields two identical grooves into which short segments of the target proteins bind. Intriguingly, no interactions have been reported with the exposed α helical surfaces, even though these are evolutionarily conserved. Initially, DYNLL/LC8 was proposed to mediate attachment of cargoes [for example, Bim (12) and swallow (21)] to the microtubule motor dynein. This would require that one binding groove in the dimer is associated with a dynein intermediate chain whereas the other binds the cargo. Although an alluring hypothesis, in mammals most cellular DYNLL/LC8 is not actually dynein-associated (22). Furthermore, studies of DYNLL/LC8 in complex with various targets has revealed that it is thermodynamically and structurally unlikely that DYNLL/LC8 could bind cargo while also bound to dynein (23, 24). Indeed, the DYNLL/LC8 dimer is now considered to have an essential role in macromolecular complex formation by driving dimerization and stabilizing inherently disordered segments of target proteins (25).

Fig. 1

Structural features of DYNLL1. (Top) The molecular surface of DYNLL1, the location of two residues (His55 and Ser88) that have been implicated in monomer-dimer transitions, and the Cys residues involved in disulfide bond formation are shown. Both His55 (H, lime), which is affected by alterations in pH, and Ser88 (S, salmon), which is phosphorylated in a PAK1-dependent manner in vivo, are located at the dimer interface. In contrast, Cys2 (C, yellow) is positioned far from that interface. Given the ~54 Å distance between Cys2 and Cys2′, it is possible that disulfide bond formation in response to an increase in ROS occurs between rather than within dimers. (Bottom) Ribbon diagrams of the DYNLL1 dimer illustrate the strand-switched β sheet interface between the two monomers. Target proteins bind to these β sheets by forming an additional β strand. The DYNLL1 structure [Protein Data Bank (PDB) identification code 1F3C (http://www.rcsb.org/pdb/explore/explore.do?structureId=1F3C)] was displayed using PyMOL (DeLano Scientific Limited Liability Corporation, San Carlos, California; http://www.pymol.org). To view an interactive version of the bottom images, see the Interactive Structure (http://stke.sciencemag.org/cgi/content/full/sigtrans;1/47/pe51/DC1).

Because only dimeric DYNLL/LC8 is apparently competent to bind target proteins, control of the monomer-dimer equilibrium potentially acts to regulate DYNLL/LC8-driven associations. In vitro, this transition may be modulated by alterations in pH, because protonation of the conserved His55 results in dimer dissociation (Fig. 1) (26). Similarly, upon epidermal growth factor (EGF) stimulation, phosphorylation of DYNLL1 at Ser88 occurs in response to activation of the kinase PAK1, and phosphorylated DYNLL1 has been proposed to enhance division of breast cancer cells (27). Further analysis of Ser88→Glu88 phophomimetic mutants indicated that this modification indeed modulates the DYNLL1 dimer-monomer transition (28, 29). However, the direct involvement of PAK1 in the phosphorylation of DYNLL1 has recently been brought into question because structural studies suggest that PAK1 binding to DYNLL1 actually obscures this putative phosphorylation site from the kinase active site and that, at least in vitro, PAK1 does not phosphorylate recombinant Drosophila DYNLL/LC8 (30).

So does DYNLL/LC8 merely function as the molecular glue that holds complex macromolecular systems together? Several lines of evidence suggest that the answer is no, that DYNLL/LC8 also has regulatory functions. Analysis of nNOS suggested that DYNLL1 might control enzyme function by regulating oligomeric status because only dimeric nNOS is enzymatically active (11). Similarly, control of dimerization of DYNLL/LC8 and thus its binding partners by phosphorylation of Ser88 may regulate DYNLL/LC8 function during tumorigenesis (27, 28). Now, Jung et al. report the association of DYNLL1 with IκBα (inhibitor of NF-κB α) (31) and that this interaction is involved in regulating NF-κB activation in response to changes in cellular redox balance (16). This effect of DYNLL1 on the NF-κB pathway provides further support for the concept that the functions of DYNLL1 extend beyond just structural contributions to include regulatory roles.

The NF-κB heterodimer (composed of p65 and p50 subunits) acts as a transcription factor and is kept inactive by sequestration in the cytoplasm by IκBα, which hides the nuclear localization signal. After phosphorylation by the IκB kinase in response to signals such as tumor necrosis factor α (TNF-α), IκBα is ubiquitinated and destroyed by the proteasome, thus allowing NF-κB to traffic into the nucleus (32). TNF-α also enhances the generation of reactive oxygen species (ROS) by NADPH (the reduced form of nicotinamide adenine dinucleotide phosphate) oxidase, which also promotes NF-κB activation (33). Jung et al. propose that DYNLL1 acts as a negative inhibitor of NF-κB activation by binding to IκBα and preventing its phosphorylation and degradation. Furthermore, DYNLL1 may provide the regulatory linkage between ROS generation and NF-κB activation. Specifically, they demonstrated that DYNLL1 is normally maintained in the reduced state by a thioredoxin-related protein (TRP-14) that is in turn kept reduced by cellular NADPH-derived reducing equivalents cycled through a thioredoxin reductase. However, once DYNLL1 transitions to the oxidized state because of enhanced ROS levels, DYNLL1 appears to release IκBα, which would then allow phosphorylation and proteolytic processing of IκBα, leading to NF-κB activation.

Each DYNLL1 monomer contains three cysteine residues that potentially could mediate disulfide formation. Mutagenesis studies by Jung et al. suggest that Cys2 is the important residue for disulfide formation; this is the only Cys that is not present in the other mammalian isoform, DYNLL2, or in nonmammalian orthologs. However, the two Cys2 residues in the crystal and nuclear magnetic resonance structures are exposed on diametrically opposed faces of the dimer and separated by ~54 Å, so that it is hard to envision how disulfide formation within a dimer might occur without an enormous and energetically highly unfavorable alteration in conformation. One possibility is that the disulfide is actually formed between two DYNLL1 dimers (interdimer) rather than within a single dimer (intradimer). Alternatively, because the dimer-monomer transition is regulated in response to both phosphorylation and pH, it is feasible that disulfide formation occurs subsequent to or coincident with dimer dissociation; this would expose the region involved in formation of the dimer interface and potentially might stabilize a state approximately equivalent to the monomer in a redox-sensitive fashion.

Of course, this model now raises the major question of how DYNLL1 function in the NF-κB pathway can be controlled in response to a global alteration in cellular redox state without dramatically affecting other systems (such as cytoplasmic dynein) that rely on DYNLL1 dimer function for activity. One possibility here, assuming a functional overlap between the two isoforms, is that there is sufficient DYNLL2 to allow these other systems to operate, which would be consistent with the lack of a lethal phenotype upon small interfering RNA–mediated reduction of DYNLL1 in HeLa cells (16). A second puzzle is that, in other systems, DYNLL/LC8 appears to promote dimerization of target proteins through its two identical binding grooves. However, in the model presented by Jung et al. it is not clear what interacts with the second target binding site on DYNLL1. Perhaps functional IκBα is actually oligomeric or there is another unidentified player in this regulatory pathway.

In conclusion, the highly conserved DYNLL/LC8 proteins act as dimerization engines in many macromolecular systems, ranging from actin- and microtubule-based motors to apoptotic factors, transcription complexes, and viral components. Furthermore, control of the monomer-dimer transition by modification of residues located at the dimer interface has now been linked to regulation of enzymatic function in several systems. The work on NF-κB activation extends the concept of DYNLL/LC8-mediated regulation to include responding to alterations in cellular redox poise. The puzzle now is how NF-κB can be controlled in this manner without disrupting the activity of other cellular systems that use the DYNLL1 dimer as an essential component.

Acknowledgments

My laboratory is supported by grant GM51293 from the NIH.

References and Notes

View Abstract

Navigate This Article