Read with caution!

This post was written during early stages of trying to understand a complex scientific problem, and we didn't get everything right. The original author no longer endorses the content of this post. It is being left online for historical reasons, but read at your own risk.

The cell has two major mechanisms for degrading proteins – the lysosomes and the proteasome.  Recently I discussed the evidence for lysosomal involvement in the degradation of PrP, including both PrPC and PrPSc.  Here I review the evidence that proteasome has a role in degrading PrP, and that PrPSc may interfere with proteasome function.


PrP is a GPI-anchored secretory pathway protein which spends most of its life on the cell surface or in endocytic vesicles, cycling between the cell surface and endosomes.  Because the proteasome exists in the cytosol and is thought to primarily degrade cytosolic proteins, it may have initially seemed non-intuitive that it could be involved in degrading PrP.  But in the late 1990s, details began to emerge about ER-associated degradation (ERAD), a mechanism by which some misfolded secretory pathway proteins are recognized in the ER and exported via retrotranslocation to be degraded by the proteasome.

It wasn’t long before evidence emerged that PrP was one such protein.  First, it was shown that Y145X PrP, a truncated anchorless mutant that causes cerebral amyloid angiopathy, is (almost) entirely degraded via ERAD [Zanusso 1999 (ft)], and that a subset of Q217R PrP molecules, a mutant causing GSS, are degraded by ERAD as well [Jin 2000 (ft)].

Next, two studies published around the same time demonstrated that ERAD wasn’t limited to mutant PrP, but also applied to wild-type PrP. [Yedidia 2001, Ma & Lindquist 2001].

Both studies started from the observation that treating cells with proteasome inhibitors causes PrP to accumulate, implying that normally PrP is degraded by the proteasome.   Yedidia used three different proteasome inhibitors (ALLN, lactacystin and MG132) and got similar results with all three; Ma & Lindquist only used MG132.

Yedidia’s experiments were conducted in CHO cells, and the 3F4 antibody was used to detect PrP.   The use of 3F4 introduces two potential confounders.  First, alpha cleavage of PrP (before or after codon 111) abolishes the 3F4 epitope (codons 106-112 inclusive), so any effects on alpha cleavage rates will affect the quantification of protein.  Second, the 3F4 epitope is buried in PrPSc, a fact which is the basis for conformation-dependent immunoassay (CDI) [Safar 1998] so if any treatment were to cause conformational changes, then that too would affect the readout.  Of course, spontaneous prion formation is rare, but PrP can adopt many conformations – as one example, copper treatment can cause a conformational change resulting in protease resistance [Quaglio 2001 (ft)].  And indeed, Yedidia found that treating cells with the proteasome inhibitor ALLN did indeed cause PrP to acquire detergent insolubility and some degree of protease resistance, two properties normally associated with PrPSc, implying that some degree of misfolding had occurred.

Perhaps the most compelling evidence for proteasomal involvement in PrP degradation from Yedidia’s paper was the presence of ubiquitylated “ladders” of PrP on Western blots.  Proteins destined for degradation by the proteasome often get tagged with more than one ubiquitin molecule at 8 kDa each, meaning that the protein will migrate on a gel at various weights separated by 8 kDa increments.  Of note, Yedidia briefly acknowledges the second confounder I raised above, stating that “PrP denaturation in hot SDS was essential for detecting the ubiquitylated bands”.

Ma & Lindquist 2001 focused on an NT-2 (human neuroblastoma) cell line but also used a couple of different cell lines – COS-1 (monkey kidney cells) and mouse NIH 3T3 fibroblasts.  They examined the behavior of endogenous wild-type PrP in the cell lines, and they also tried transfecting them with PrP cDNA constructs to achieve expression of two PrP sequences of interest.  Specifically, they used two cDNA constructs.  One was an essentially wild-type sequence of mouse PrP but with the (human) 3F4 epitope knocked in (109M and 112M) for ease of detection with the 3F4 antibody.  The other construct was MoPrP with 3F4 and the D177N mutation – equivalent to the D178N FFI mutation in humans.  I say FFI, not 178 fCJD, since MoPrP is always 128M (129M in human numbering).  As far as I know, these were the same constructs later used to create the fatal familial insomnia knock-in mice and their controls: ki-3F4-FFI and ki-3F4-WT [Jackson 2009].

In the untransfected cells, Ma & Lindquist found that treating the cells with proteasome inhibitors caused PrP to accumulate.  Various stains to look for co-localization of PrP with markers of different subcellular compartments suggested that this accumulating PrP was indeed in the cytoplasm, a conclusion which Yedidia had reached as well.  The presence of PrP in the cytosol raises two possibilities.  Either this PrP had failed to be cotranslationally translocated and had been spit out by a ribosome directly into the cytosol, or it had been translocated into the ER and then retrotranslocated back out into the cytosol.  PrP that had seen the inside of the ER would be expected to have undergone post-translational modifications and thus lack the N-terminal signal peptide and C-terminal GPI anchor signal sequence, whereas PrP that never got translocated to start with would be expected to contain all 253 (254 in MoPrP) amino acids. That’s easy enough to test. On Western blots, the accumulated PrP in cells treated with proteasome inhibitors turned out to have the same molecular weight as regular ER-processed PrP.  Could that be a coincidence whereby it represented a cleaved form of a non-post-translationally-modified PrP? No, since it still reacted with antibodies to the N- and C-termini of the mature, ER-processed protein.  This strongly suggested that PrP was reaching the cytoplasm via retrotranslocation.

When the same cells were transfected with the PrP cDNA constructs bearing the FFI mutation, PrP was seen to accumulate in the cytosol even without proteasome inhibitors.  Meanwhile, less of this FFI PrP ever reached the cell surface than the PrP from the wild-type construct did.  This suggested that the FFI PrP was misfolding very early in its life, in the ER, and that the cell has a mechanism for recognizing misfolded PrP in the ER and retrotranslocating it for proteasomal degradation.  In turn, the fact that even wild-type endogenous PrP did accumulate to detectable levels in the cytoplasm upon treatment with proteasome inhibitors suggested that in healthy wild-type cells, some misfolded PrP is indeed produced all the time, but at a rate that the cell is able to keep pace with.

It was an elegant story: cells could keep pace with a low, constant level of spontaneous misfolding of PrP as long as their proteasomes were healthy, but had more trouble keeping pace with the higher misfolding rate of FFI PrP.  Given that FFI PrP was reduced (compared to wild-type) at the cell surface but present in the cytosol, one might ask whether total PrP levels were higher or lower in FFI cells – Ma & Lindquist don’t address this.  Certainly, the FFI mice created with the same constructs years later appear to have reduced PrP protein levels compared to wild-type [Jackson 2009].  Curiously, Ma & Lindquist did not cite earlier work suggesting that the unglycosylated forms of FFI (as well as D178N 129V) PrP are degraded before reaching the cell surface [Petersen 1996].  Petersen’s evidence had pointed to lysosomal degradation, rather than proteasomal degradation.

Between Yedidia and Ma & Lindquist’s work, it was looking pretty clear that PrP could normally be degraded by the proteasome, but another work a couple of years later questioned this conclusion [Drisaldi 2003 (ft)].  Drisaldi sought to confirm the accumulation of PrP upon proteasome inhibition by a different approach: rather than just blotting for total PrP, Drisaldi did a pulse-chase experiment to determine the half-life of PrP in the presence and absence of proteasome inhibitors, and found no difference.  Drisaldi did see PrP accumulating after proteasome inhibitor treatment, just as Yedidia and Ma & Lindquist had, but since PrP’s half-life didn’t increase, the accumulation didn’t appear to be due to an decreased degradation rate.

Instead, Drisaldi found a few different artifacts which could had been potential confounders in the earlier studies.  First, one type of mutant PrP (PG14, carrying a nine octapeptide repeat insertion) passaged more slowly than wild-type PrP through the endoplasmic reticulum – if this were true of FFI PrP as well, it could help to explain why Ma & Lindquist saw reduced cell surface FFI PrP.  Curiously, Drisaldi did not report whether this was true of FFI PrP, even though Drisaldi did use FFI PrP in other experiments.

Second, Drisaldi found that long-term ( > 8h) treatment of cells with proteasome inhibitors caused PrP mRNA to increase – in retrospect, conceivably a result of PrP transcriptional activation in response to ER stress [Dery 2013 (ft)].  Therefore, although Drisaldi did see an increase in PrP protein in cells treated with proteasome inhibitors just as previous authors had, Drisaldi did not consider this to mean that PrP was constitutively degraded by the proteasome.  Indeed, Drisaldi used an antibody against PrP’s signal peptide and found that cells transfected with PrP cDNAs tended to produce a cytosolic form of PrP which bore an intact signal peptide.  This suggested that no ERAD was occurring but, rather, that at artificially high rates of PrP synthesis not all PrP was making its way into the ER in the first place.

However, Drisaldi’s results cannot simply explain away all of the previous findings.  Notably, if PrP is not degraded by the proteasome, why did Yedidia find ladders of polyubiquitylated PrP?  And if all of the cytosolic PrP in transfected cells was just never translocated to the ER in the first place, why did Ma & Lindquist find a form of cytosolic PrP whose molecular weight and antibody reactivity perfectly matched a version of PrP with the signal peptide and GPI signal chopped off?

Looking at all three studies together, it still seems likely that the proteasome does play a role in constitutive degradation of PrP and that at least some of the increase in PrP seen in Yedidia and Ma & Lindquist’s studies was genuinely due to decreased degradation rates.  At the same time, it seems likely that some PrP (at least in transfected cells) can be translocated directly into the cytosol – the band for the signal peptide on Drisaldi’s Western blots is quite clear – and that PrP mRNA may increase under some experimental conditions, an important confounder to be borne in mind.

All three of these studies looked strictly at uninfected cells and so, although PrP mutants were examined, the case for a role of proteasomal inhibition in actual prion disease had yet to be made.  A few years later, another study made a convincing case that PrPSc oligomers do inhibit the proteasome and that this may contribute to prion pathology [Kristiansen 2007 (ft)].

Kristiansen compared infected and uninfected mouse neuronal cell lines (GT-1 and PK-1, the latter a subclone of N2a) and found reduced proteasomal activities in the infected lines.  And some proportion of the PrPSc in the infected cells was found to colocalize with cytosolic markers.  Though these neuroblastoma cell lines are far from normal – N2a have been in continuous culture for over 50 years – these were at least nominally wild-type cells, untransfected and untreated with proteasome inhibitors, and thus the confounders raised by Drisaldi are unlikely to apply here.   Kristiansen also used a reporter vector expressing ubiquitylated GFP (UbG76V-GFP).  This protein is supposed to be degraded by the proteasome, so if it accumulates instead of being degraded, that means the proteasome is probably inhibited – and indeed, the proteasomes in the infected cell lines appeared inhibited.  Similar results were obtained with a synthetic beta sheet rich form of PrP.

Kristiansen presented a variety of other evidence narrowing down the specific mechanisms of proteasome inhibition, and a subsequent work by the same group has determined that PrPSc oligomers stabilize the 20S proteasome’s entry channel in a “closed” configuration [Deriziotis 2011]

The most compelling part of the paper is the in vivo evidence.  Fascinatingly, there are transgenic mice that constitutively express UbG76V-GFP, such that GFP accumulates in the brains of these mice if and only if the proteasome is inhibited [Lindsten 2003].  Kristiansen infected some of these mice with 22L prions and found that infected mice exhibited an accumulation of GFP by the terminal stages of illness, with the amount of GFP correlating with the extent of neuropathology, while uninfected mice had no GFP accumulation.

Kristiansen’s work strongly supports the notion that proteasomal inhibition by PrPSc really does occur during prion disease.  The authors suggest that proteasome inhibition may account for some of the pathology of prion disease and contribute to neuronal death.

relevance to therapeutics

There are two potential implications for therapeutics development here.

First, if PrP is constitutively subject to some degree of proteasomal degradation via ERAD, hastening this degradation could be one therapeutic route towards lowering PrPC.  Unfortunately, while we have plenty of drugs that induce autophagy and could promote lysosomal degradation of PrPC, I’m not aware of any drugs that increase proteasomal function.

Second, if some of the neurotoxicity of prion disease owes to PrPSc oligomers inhibiting the proteasome, then alleviating this inhibition could be a therapeutic goal as well.

Jeff Kelly, inventor of Tafamidis, has been on tour for a while now promoting the idea that we’re unlikely to find separate small molecule treatments for every single different neurodegenerative disease, but rather should invest our energies in treatments that will promote the degradation of misfolded proteins generally.  Whether this could work for prion disease is an open question.  One concern is that prion disease is about an order of magnitude more rapid than any other neurodegenerative disease,  measured in months rather than years – could a non-specific treatment make much difference in the face of such a rapid accumulation of misfolded PrP?  Another concern is whether PrP obeys enough of the same regulatory mechanisms as other amyloidogenic proteins to be treatable by the same means.  As one example, we know that misfolded protein accumulation in the ER triggers phosphorylation of PERK and ultimately eIF2α, reducing translation rates of all proteins globally.  In theory, this should relieve pressure on the cell’s protein-folding machinery – but PrP translation rates are paradoxically increased upon eIF2α phosphorylation [Moreno 2012].  Salubrinal, a compound which promotes eIF2α phosphorylation, should theoretically help cells cope with misfolded protein stress but instead actually hastened the onset of terminal illness in Moreno’s prion-infected mice.

While activating the proteasome or freeing it from inhibition by PrPSc may be a potential therapeutic route, the road forward isn’t very clear yet.