Cytoskeleton grows p62 condensates for autophagy

p62 body, a representative selective cargo for autophagy, is known to be formed via liquid–liquid phase separation, but its spatiotemporal regulation remains largely unknown. In a recent paper published in Cell Research, Feng et al. show that branched actin filaments and Myo1D collect stochastically formed nanoscale p62 bodies and grow them into micron-scale condensate, which is prerequisite for their autophagic degradation.

Liquid–liquid phase separation (LLPS) of biomolecules is the main mechanism for organizing various membraneless organelles in the cytoplasm and nucleoplasm and plays crucial roles in diverse cellular processes.¹ One such process is autophagy, a lysosomal degradation system of cytoplasmic materials, in which both the autophagy machinery and the selective autophagy substrates have been reported to be regulated by LLPS.^2,3 p62 is a well-established selective autophagy receptor for ubiquitinated proteins and is known to form p62 bodies together with poly-ubiquitinated proteins, which are selectively degraded by autophagy. Recent studies revealed that p62 undergoes LLPS together with poly-ubiquitin chains through multivalent interactions in vitro, which is proposed to be the mechanism of p62 body formation in cells.^4,5 However, the molecular mechanisms of how p62 body formation is regulated in cells is largely unknown.

In a recent paper published in Cell Research, Feng et al. studied the role of cytoskeletal elements in the formation and autophagic degradation of p62 body.⁶ They first studied the effect of pharmacological reagents perturbing various cytoskeletons on puromycin-induced p62 body formation and found that perturbation of Arp2/3-derived branched actin network most severely prevented p62 body formation without affecting the levels of p62 and ubiquitinated proteins. Fluorescence microscopy showed that p62 and actin filaments colocalized to each other and p62 bodies appeared throughout the branched actin network, which was further confirmed by a peroxidase-mediated proximity labeling technique coupled with transmission electron microscopy (APEX2-TEM). Arpc2 and Arpc1b, two subunits of the branched actin nucleator Arp2/3 complex, colocalized with p62 bodies and knockdown of Arpc2 severely compromised the formation of p62 bodies. From these data, the authors concluded that Arp2/3 complex and branched actin networks are required for the formation of p62 bodies.

The authors next examined the role of actin-associated motor proteins in the LLPS of p62. Among six homologs of type I myosin, they found that Myo1D was most well colocalized with p62 bodies. Colocalization of Myo1D with p62 bodies was further confirmed using various methods that include structured illumination microscopy, APEX2-TEM, and correlative light electron microscopy. Fluorescence recovery after photobleaching experiments revealed that Myo1D within p62 bodies dynamically exchanged with that in the cytosol, indicating the liquid-like feature of p62 bodies. Incorporation of Myo1D into p62 bodies and their dynamic feature were also confirmed by in vitro reconstitution experiments. Myo1D is composed of head, neck, and tail domains, and domain analyses revealed that the neck and tail domains physically interact with poly-ubiquitin/p62 complexes, suggesting that these physical interactions promote incorporation of Myo1D into p62 bodies.

The authors then studied the role of Myo1D in p62 body formation. Myo1D knockout (KO) did not affect the formation of nanoscale p62 bodies, but severely impaired formation of micron-scale p62 bodies. To access the mechanism of how Myo1D participates in the formation of micron-scale p62 bodies, the authors performed mutational analysis of Myo1D. Although the truncation mutants that lack each domain could partition into nanoscale p62 bodies, they failed to induce coalescence of these small p62 bodies into micron-scale bodies. Importantly, a single mutation (K108R), which impairs the motor function of Myo1D, failed to rescue the KO effect, indicating that the motor activity of Myo1D is required for the micron-scale p62 body formation in cells. The authors then established an in vitro reconstitution system on the planar cover glass that contains an Arp2/3-derived branched actin network and p62 bodies composed of p62 and poly-ubiquitin chains. After addition of Myo1D, micron-scale p62 bodies were observed on the reconstituted branched actin network. Myo1D colocalized with the network irrespective of p62. Moreover, micron-scale p62 body formation on the branched actin network was dependent on ATP, which was impaired by the K108R mutation. Since Myo1D alone did not accelerate p62 body formation, these data suggested that Myo1D utilizes its motor activity to promote p62 body formation by collecting p62 locally.

The authors finally studied the significance of the formation of large micron-scale p62 bodies for their autophagic degradation. An autophagy machinery protein LC3 was well colocalized with micron-scale p62 bodies, but much less frequently with nanoscale p62 bodies. Moreover, knockdown of Arpc2 or Myo1D, which reduces the size of p62 bodies in cells, markedly attenuated autophagic degradation of p62. These data suggested that the sufficient size (micron-scale) of p62 bodies is important for their autophagic degradation.

This study, focusing on p62 LLPS, proposed an attractive concept that cytoskeletal dynamics promote LLPS locally within cells (Fig. 1). This process required not only branched actin network, but also the motor activity of Myo1D and ATP, indicating that Myo1D would bring p62 along the actin network to branched regions, which would increase the local concentration and grow p62 bodies. In the presented data, the authors have not succeeded in observing movement of p62 along the actin network in vivo and in vitro, and thus have not demonstrated that Myo1D moves either nanoscale p62 bodies or p62 polymers, or both. These observations would markedly increase our understanding of the role of cytoskeletal dynamics in regulation of p62 LLPS. Another important future issue is to know how general the regulation of LLPS by cytoskeletal dynamics is utilized in proteins other than p62.

Fig. 1: Proposed model of micron-scale p62 body formation.

p62 polymerizes via the PB1 domain and binds to ubiquitin using the UBA domain. Multivalent interaction between p62 and poly-ubiquitin chains induces their LLPS to form nanoscale condensates, which are then delivered along the branched actin filament via Myo1D and grow into micron-scale condensates through coalescence.

This study also proposed another important concept that the sufficient size of condensates is important for their selective degradation by autophagy. Thus far, the liquidity of protein condensates was shown to be important for their autophagic degradation,^7,8 but their size was scarcely studied. Recently, p62 bodies were shown to function as a site of autophagosome formation and thereby promote selective autophagy of themselves.^9,10 Micron-scale size might be required for condensates to function as the autophagosome formation site, which should be validated in the future analysis.

In conclusion, this study proposed an important concept that cytoskeletal dynamics regulate LLPS of a cytosolic protein, which is expected to accelerate mechanistic studies on the spatiotemporal control of LLPS of diverse proteins in cells.