Write a comment

In this article, the authors convincingly demonstrate that the ubiquitin E3 ligase HOIL-1 can ubiquitylate in vitro oligosaccharides, only the second example ever reported of a non-proteinaceous ubiquitylation substrate. They also observe that mice harbouring a catalytic dead version of HOIL-1 cannot efficiently clear polyglucosan from relevant tissues, thereby hypothesising that this aberrant phenotype may arise due to such inclusions not being removed in a HOIL-1 dependent manner. The presented in vitro data are robust, stemming from meticulously-thought, well-controlled experiments. The in vivo observations are interesting, with fascinating appeal should the hypothetical model the authors put forward be true. Having said so, it is surprising to me that the authors have not validated some of their in vitro findings or investigated potential functional implications in in vivo settings. This is my main point of concern, on which I elaborate subsequently, providing a few experimental suggestions that may help address such potential criticism that may come up in the input of other reviewers. I also provide feedback with respect to some additional controls and minor technical issues that may help cover all bases.

Though the in vitro data of HOIL-1 dependent ubiquitylation of glycogen and maltoheptaose are abundantly clear, there is no evidence that this occurs in cells. Furthermore, even if HOIL-1 indeed ubiquitylates those biomolecules in vivo, the mechanism (monoubiquitylation? multi-monoubiquitylation? en bloc polyubiquitylation? chains of what linkage? what proportion of the a1:4 gets ubiquitylated?) and the functional implications (different glycophagy rates? different lysosomal activity? how is the lack of a1:6 signalled?) are still completely unclear.

I think that any in vivo/cell data that (in)directly corroborate their in vitro findings would substantially increase the impact of and interest in their study. Here are some experiments that may help the authors further validate their conclusions:

1. The authors could repeat the experiments carried out in Figure 2 and/or 4, using HOIL-1 containing complexes isolated (immnoprecipitated) from wt or C458S cells instead of recombinant proteins, to demonstrate that endogenous HOIL-1 complexes recapitulate the in vitro results, namely that they ubiquitylate recombinant glycogen and directly interact with oligosaccharides.
2. Using a non-basic purification method, the authors could try to isolate glycogen from wt or C458S cells and investigate its ubiquitylation status either by blotting or mass spectrometry.
3. To functionally connect the phenotype of aberrant glycogen to the lack of HOIL-1 ubiquitylation activity with actual in vivo implications, the authors could test what happens to glycophagy in wt versus C458S cells. Do C458S cells exhibit increased authophagy markers? Do those markers colocalise with PAS staining at those cells? Does the expression of lysosomal a1:4 glycosylase or Stbd1 increase in C458S cells? Do those proteins increasingly colocalise with PAS staining at those cells? Can you rescue the C458S phenotypes of aberrant glycogen deposits by overexpressing Stdb1 or another way of artificially stimulating glycophagy?
4. The authors could exogenously produce glycogen lacking a1:6 junctions and put it in cells. Would it recapitulate the HOIL-1 (C458S) phenotypes? Would the HOIL-1 (C458S) be less susceptible to such a substance?

The authors claim in their discussion that HOIP-produced M1 oligomers may be powerful allosteric activators of HOIL-1 ubiquitylation of glycogen. That is just one of many possibilities. Trying to exclude other possibilities, the authors could test the effect of other ubiquitin polymers (e.g. K48- or K11-polyubiquitin tetramers) on recombinant HOIL-1 reactions. They could also investigate if cells exclusively expressing catalytically inactive HOIP exhibit similar aberrant glycogen structures as HOIL-1 (C448S) cells do. If so, are the HOIP-CS/HOIL-1 CS phenotypes epistatic?

Furthemore, the authors could investigate if other IBR E3 ligases (e.g. HHAR1/ARIH1 or Parkin) can in vitro ubiquitylate maltoheptaose or if this is a property unique to HOIL-1. Such data would complement well their current data exhibiting that other IBRs do not bind oligosaccharides in vitro.

In my opinion, the authors would benefit by staining for specific molecules (e.g. by western blotting or PAS staining), in parallel to their coomassie stainings, particularly in Figures 5-7 where differences based solely on molecular weight are sometimes quite hard to discern.

Some minor points:

1. In figure 1, the authors could show PES stainings of tissues that do not show any increase, as a negative control.
2. In figure 2, the authors could treat ubiquitylated glycogen with recombinant USP2 to further validate the fact that this is a ubiquitylation event.
3. Though panels 4F, 6G and 6H are informative, they are not representing actual data but theoretical projections, as such I think that they would be more suited for supplementary information.
4. Panels 5B and 5C/5D are somewhat at odds with one another with respect to the remaining, unmodified maltoheptaose. Though the levels of modified maltoheptaose increase similarly, the levels of unmodified maltoheptaose are visibly higher in the NZF mutants/deletion when ubiquitin tetramers are used.