Publications

Comparison of cryo-EM-based atomic models of myosin-free and myosin-decorated actin-tropomyosin filaments shows a clear azimuthal shift in tropomyosin position between corresponding C- and M-state thin filaments. Arg 369 and Glu 370 on the tip of the Loop-4 of the myosin-head likely drive C-state tropomyosin to its M-state position, while Loop-4 along with the cardiomyopathy (CM) motif and other surface loops on the myosin-head elicit the tight binding characterizing the M-state myosin-head interaction with actin. These cryo-EM-based models of myosin-decorated thin filaments were generated from filaments saturated with myosin-S1, i.e., here all actin subunits along thin filaments contain a bound myosin-head. However, this arrangement is non-physiological, since in actively contracting muscle, few myosin-heads attach to neighboring actin subunits along thin filaments at the same time. Thus, the collective effect of multiple interacting myosin-heads on the positioning of tropomyosin may be amplified artificially over what is physiologic. In the current cryo-EM study, we controlled the level of S1-binding to actin-tropomyosin. We selected and analyzed filament segments where only a single myosin-head bound to stretches of myosin-free actin-tropomyosin. The single myosin-heads and associated actin-tropomyosin were successfully aligned to each other yielding a 5 Å resolution 3D-reconstruction. Tropomyosin on the actin helical strand containing the bound myosin-head occupied the M-state position, which was found to be indistinguishable from the position of tropomyosin in fully decorated filaments. In contrast, tropomyosin, on the opposite side of double-helical actin that is “untouched” by myosin, occupied the C-state position. Our new precision will allow us to determine structurally the cooperative unit size for the myosin-induced C- to M-state tropomyosin transition and to investigate factors that may lead to cross-strand tropomyosin communication.

Electrostatic interactions between actin residues K326 and K328 and tropomyosin bias tropomyosin to an F-actin location where it blocks myosin attachment. K326/328 acetylation neutralizes their charge, potentially disrupting thin filament-based contractile regulation. We verified acetylation of K326/328 on human cardiac actin (ACTC1) and generated recombinant K326/328Q, pseudo-acetylated ACTC1. Pseudo-acetylation reduced inhibition of myosin-driven motility of F-actin-tropomyosin and F-actin-tropomyosin-troponin at low Ca2+. Cryo-EM-based and computational modeling revealed that pseudo-acetylation did not alter tropomyosin positioning along F-actin but decreased local F-actin-tropomyosin interaction energy. Thus, by reducing the energetic demands required for myosin to displace tropomyosin, ACTC1 K326/328 acetylation may promote contractile activation.

In light chain amyloidosis (AL), aberrant monoclonal antibody light chains (LCs) deposit in vital organs causing organ damage. Each AL patient features a unique LC; previous cryogenic electron microscopy (cryo-EM) studies revealed different amyloid structures in different AL patients. How LC mutations influence amyloid structures remains unclear. We report a cryo-EM structure of cardiac AL-224L amyloid (2.92 Å resolution) from λ6-LC family, which is overrepresented in AL amyloidosis. Comparison with λ6-LC structures from two other patients reveals similarities in amyloid folds, along with major differences caused by specific mutations. Differences in AL-224L include altered C-terminal conformation with an exposed surface forming an apparent ligand-binding site; an enlarged hydrophilic pore with orphan density; and altered steric zipper registry with backbone flipping, which likely represent general adaptive mechanisms in amyloids. The results reveal shared features in λ6-LC amyloid folds and suggest how mutation-induced structural changes influence amyloid-ligand interactions in a patient-specific manner.

Epigenetic macromolecular enzyme complexes tightly regulate gene expression at the chromatin level and have recently been found to colocalize with RNA splicing machinery during active transcription; however, the precise functional consequences of these interactions are uncertain. Here, we identify unique interactions of the CoREST repressor complex (LSD1-HDAC1-CoREST) with components of the RNA splicing machinery and their functional consequences in tumorigenesis. Using mass spectrometry, in vivo binding assays, and cryo-EM, we find that CoREST complex–splicing factor interactions are direct and perturbed by the CoREST complex selective inhibitor, corin, leading to extensive changes in RNA splicing in melanoma and other malignancies. Moreover, these corin-induced splicing changes are shown to promote global effects on oncogenic and survival-associated splice variants, leading to a tumor-suppressive phenotype. Using machine learning models, MHC IP-MS, and ELISpot assays, we identify thousands of neopeptides derived from unannotated splice sites that generate corin-induced splice-neoantigens that are demonstrated to be immunogenic in vitro. Corin is further shown to reactivate the response to immune checkpoint blockade, effectively sensitizing tumors to anti–PD-1 immunotherapy. These data position CoREST complex inhibition as a unique therapeutic opportunity that perturbs oncogenic splicing programs while also creating tumor-associated neoantigens that enhance the immunogenicity of current therapeutics.