The "lower dimer" and its role in actin patterning: studying different forms of actin by electron microscopy, biochemistry and tailor-made antibodies

At the onset of actin polymerization, a dimeric actin, the so-called ‘lower dimer’ (LD), is transiently incorporated into growing filaments and subsequently dissociates. Because of the antiparallel orientation of the two actin subunits, the LD conformation is incompatible with the helical geometry of F-actin filaments and thus, LD- contacts are not present in conventional F-actin filaments. By copolymerizing G-actin and previously cross-linked LD in the presence of phalloidin, we produced a novel type of actin hybrid filaments (HF) where the LD persists even at steady state. Biochemical analysis and electron microscopy (EM) showed that copolymerization of G-actin with increasing amounts of LD yielded HF with incorporated LD at corresponding ratios. In the HF, one of the LD subunits constituted the filament backbone, whereas the other subunit jutted out from the HF surface, causing lateral protrusions and filament branches. Locally unraveled HF strands and irregular crossover spacings indicated that LD incorporation interfered with interstrand, and possibly intrastrand subunit contacts. In highly disordered filament sections, the subunit contacts needed for phalloidin binding were most likely not established. Surprisingly, scanning transmission electron microscopy (STEM) data did not reveal any significant difference between mass-per-length and full-width half-maximum in HF versus conventional F-actin filaments. Our findings corroborate the intrinsic ability of actin for branching in vitro, and that the LD plays a role in supramolecular actin patterning.
We next tested the immunogenic potential of a novel type of peptide nanoparticles (NP). 60 copies of a computationally designed and recombinantly expressed coiled coil polypeptide self-assemble to form regular dodecahedrons that represent a repetitive antigen display platform. By engineering either a 13mer corresponding to aa 239 - 251 of either wild-type human beta-cytoplasmic actin or containing a point mutation at aa 245 onto the NP, we obtained two homogenous populations of NP. Proof of concept was provided by immunizing rabbits with NP and producing antisera that reacted with wild type and mutant actin, respectively. A prominent structural element, the ‘hydrophobic loop’ (h-loop), is exposed to the surface of G-actin, but buried within F-actin filaments. We assumed that immunization with NP exposing the h-loop would yield antibodies, which detect only non-filamentous actin. Depending on the constructs chosen, the NP polypeptides expressing 12-mers corresponding to the h-loop of beta-actin assembled into disperse spherical particles or aggregated extensively. Monoclonal antibodies (mAb) raised against aggregated NP specifically reacted with the respective NP, purified skeletal muscle actin and actin in cell extracts under native and denaturing conditions. Moreover, immunofluorescence (IF) and immuno-EM revealed that these antibodies recognized the h-loop of actin in its native context throughout non-filamentous actin in the cytoplasm and in the nucleus. Consistently, the mAb did not cosediment with purified skeletal muscle F-actin filaments at high speed. Our data shows that self-assembling peptide NP are suitable for the immunogenic surface display of actin-related epitopes that are otherwise poor antigens. Presenting the antigenic determinants in a repetitive array significantly increased the immune response in hosts, resulting in specific poly- and monoclonal antibodies.
Last, we addressed the interaction of actin with the ubiquitously expressed hnRNP raver1 (R1), which translocates during skeletal muscle differentiation into sarcomers where it colocalizes with actin related ligands. IF and immuno-EM using three mAb that recognize distinct epitopes at both termini of raver1 showed that cytoplasmic R1 is an integral part of the sarcomers of heart and smooth muscle. It mainly concentrates at sarcomeric I-Z-I bands in isolated myofibrils and in skeletal muscle. Nuclear R1 is present in all muscle types tested. In isolated Xenopus nuclei, R1 mainly locates at the distal ring of the nuclear pore complex, where actin has also been described. Preliminary data revealed that recombinant full-length R1 and a C-terminal deletion fragment induce F-actin filament bundling and suggest that R1 interacts with monomeric actin. Due to the localization of raver1 and actin in the cytoplasm and in the nucleus, one might assume that the two proteins translocate as complexes to the cytoplasm, possibly accompanied by additional proteinaceous cargo.