Protein degradation by the proteasome is central to cellular homeostasis and has been studied extensively using biochemical and structural studies. Despite an in-depth understanding of core proteolytic activity, it has remained largely unresolved how individual proteasomes process substrates inside living cells where many substrate types and co-factors exist. Here, we establish a live-cell single-molecule imaging approach that enables direct visualization and quantification of protein degradation by individual proteasomes. Using this approach, we find that substrate identity, folding and protein-protein interaction have a surprisingly modest impact on processing efficiency, whereas the mode of substrate engagement greatly impacts substrate processing; degradation initiated from protein termini typically proceeds rapidly and with high processivity, whereas internal engagement constitutes a distinct processing mode that exhibits poor processivity and a specific requirement for the AAA+ family ATPase p97/VCP. Furthermore, degradation initiated from opposite termini proceeds with asymmetric rates in a sequence-dependent manner, demonstrating that directionality is an important feature of proteasomal processing in vivo. Notably, poly-glutamine substrates associated with neurodegenerative disease are efficiently degraded from one terminus but resist degradation when engaged from the opposite terminus, highlighting the importance of substrate engagement mode. Together, our results show that different modes of substrate engagement lead to different proteasomal processing outcomes in vivo and revise the prevailing view of the proteasome as a uniform degradation machine. ### Competing Interest Statement The authors have declared no competing interest.

GTSF1 is an essential activating cofactor for PIWI proteins in many metazoans. In the nematode Caenorhabditis elegans, however, GTSF-1 does not bind PIWI, but associates with the RNA-dependent RNA pol...

Transcription from CpG island (CGI) promoters controls the expression of two-thirds of mammalian genes, yet despite their prevalence, it remains unknown whether CGI-specific co-activators with intrins...

Since it was unveiled in 2020, Google DeepMind's game-changing AI tool has helped researchers all over the world to predict the 3D structures of hundreds of millions of proteins.

Protein-protein interactions (PPIs) drive virtually all biological processes, yet most PPIs have not been identified and even more remain structurally unresolved. We developed a two-step computational...

Iain Cheeseman and colleagues reveal the underappreciated role of single genes producing multiple proteins in atypical presentations of rare disease, and present case studies of affected patients thro...