Regulation of Blos1 by IRE1 prevents the accumulation of Huntingtin protein aggregates

INTRODUCTION

Protein aggregation underlies several common and devasting neurodegenerative diseases, for which there are no cures (Arrasate and Finkbeiner, 2012; Gan et al., 2018). However, the distinctive protein deposits that form within the brain are not always correlated with the specific neurons whose death or loss of function leads to the symptoms experienced by patients (Tompkins and Hill, 1997; Gutekunst et al., 1999; Kuemmerle et al., 1999). This observation has led to the idea that smaller, oligomeric forms of aggregation-prone proteins are more likely to be toxic, by interfering with normal cellular processes (Ross and Poirier, 2005). In Huntington’s disease, for example, large aggregates of the mutant Huntingtin protein are thought to be somewhat protective (Saudou et al., 1998; Arrasate et al., 2004). These aggregates can be degraded by macroautophagy (MA), which encloses aggregates in double-membraned autophagosomes that then fuse with lysosomes for degradation (Iwata et al., 2005a,b; Olzmann et al., 2008). Whether smaller, potentially more toxic, oligomers are also degraded by MA or by distinct pathways is not clear.

Aggregating proteins such as Huntingtin tend to accumulate near the juxtanuclear microtubule organizing center (MTOC) as a result of minus end-directed trafficking on microtubules, a process that is dependent on the ubiquitin-binding adapter protein HDAC6 (Kawaguchi et al., 2003; Iwata et al., 2005b; Olzmann et al., 2008). Disruption in trafficking hinders both the coalescence into larger aggregates and the degradation of these proteins (Kawaguchi et al., 2003). Minus end–directed trafficking of lysosomes has also been observed in cellular models of Huntington’s and Parkinson’s diseases (Dodson et al., 2012; Erie et al., 2015), although long-distance trafficking along axons is often disrupted in these diseases (Lie and Nixon, 2019).

We have found that clustering of lysosomes and late endosomes (LEs) near the MTOC can result from a specific pathway induced by stress in the endoplasmic reticulum (ER; Bae et al., 2019). Inositol requiring enzyme 1 (IRE1), a conserved sensor of ER stress and key mediator of the unfolded protein response, is a nuclease that cleaves the mRNA encoding the transcription factor XBP1 to initiate its splicing and activation, leading to the upregulation of many genes involved in ER homeostasis. IRE1 also cleaves several other mRNAs, initiating their degradation through the Regulated IRE1-Dependent Decay (RIDD) pathway (Hollien et al., 2009; Moore and Hollien, 2015). One of the key targets of RIDD is the mRNA encoding biogenesis of lysosome-related organelles complex 1 subunit 1 (BLOC1S1, or BLOS1). BLOS1 is a member of the BLOC1-related complex (BORC), which links lysosomes to kinesin, allowing their trafficking toward the periphery of the cell (Pu et al., 2015). BLOS1 is also important for several other trafficking and endosome-related functions, including the sorting of proteins in endosomal compartments (Zhang et al., 2014), kinesin switching during endosome recycling (Zhang et al., 2020), autolysosomal tubulation (Wu et al., 2021), and tubulin acetylation (Wu et al., 2018). Degradation of the Blos1 mRNA by RIDD during ER stress leads to the repositioning of LE/ lysosomes to the MTOC and to the enhanced degradation of ubiquitinated proteins (Bae et al., 2019).

Because neurodegenerative diseases such as Huntington’s disease often display markers of ER stress (Hetz and Saxena, 2017; Shacham et al., 2019), we hypothesized that the degradation of Blos1 may contribute to clearance of disease-associated aggregating proteins by trafficking degradative organelles to the cell center. Here we show that expression of the aggregation-prone Huntingtin protein induces degradation of Blos1 mRNA, which in turn protects cells from apoptosis and prevents the accumulation of aggregates by enhancing an alternative pathway to degradation.

RESULTS

To test whether expression of the mutant Huntingtin protein is affected by IRE1 and the RIDD pathway, we transfected mouse MC3T3-E1 cells with plasmids encoding exon 1 of either the wild-type Huntingtin protein (wtHTT), which contains a string of 23 Gln residues, or a disease-causing mutant Huntingtin (mHTT), which contains 145 Gln’s, each tagged with GFP. Exon 1 has previously been reported to recapitulate the aggregation of the full-length protein, which is often cleaved to release N-terminal fragments found in aggregates during disease progression and cellular models of Huntington’s disease (Mangiarini et al., 1996; DiFiglia et al., 1997; Wellington et al., 2000). Cells expressing wtHTT-GFP displayed a diffuse cytosolic expression pattern, whereas ∼ 20% of cells expressing mHTT-GFP displayed large juxtanuclear aggregates, as expected (Figure 1A). These aggregates colocalized with the microtubule organizing center in 66% of the aggregate-containing cells (Figure 1A). We then constructed stable MC3T3-E1 cell lines expressing the wtHTT-GFP or mHTT-GFP proteins under the control of a doxycycline-inducible promoter. We transfected HTT plasmids into cell lines expressing either Rfp (as a control) or a stabilized version of the Blos1 mRNA (Blos1^s), which contains a silent point mutation rendering it resistant to degradation by RIDD (Moore and Hollien, 2015). We previously showed that these Blos1^s -expressing cells override the repositioning of LE/lysosomes during ER stress (Bae et al., 2019). Addition of doxycycline (4.5 µM, 72 h) to control cells and the subsequent expression of mHTT, but not wtHTT, led to the activation of IRE1 as assessed by Xbp1 splicing (Figure 1B), degradation of RIDD targets (Figure 1, C and D), and repositioning of LE/lysosomes to the MTOC (Figure 1, E and F). When we induced mHTT expression in cells expressing Blos1^s, the mHTT mRNA levels were similar to that in control cells (Figure 3E) and IRE1 was activated, but LE/lysosomes did not reposition (Figure 1, B–H).

Large aggregates of mHTT were visible in only 13% of control cells in these experiments (Figure 2B), suggesting that extensive aggregation is not needed to activate IRE1. Furthermore, when we used more mild induction conditions (0.9 µM doxycycline, 36 h) such that the mHTT-GFP fluorescence remained diffuse (Figure 2, A and B), cells still degraded Blos1 mRNA and repositioned LE/lysosomes (Figure 1, G and H). This is consistent with previous studies indicating that mHTT activates ER stress pathways by interfering with the retrotranslocation and proteasomal degradation of misfolded proteins from the ER (Duennwald and Lindquist, 2008; Leitman et al., 2013), rather than through a mechanism dependent on extensive aggregation.

Several lines of evidence suggest that RIDD of Blos1 is important for degrading mHTT and avoiding the excessive accumulation of aggregates. Cells co-expressing Blos1^s were much more likely to contain juxtanuclear mHTT aggregates, and these aggregates were larger than those in control cells (Figure 2, A–C). To determine the extent of mHTT accumulation in these cells, we measured the fluorescence of its GFP tag by flow cytometry. Blos1^s-expressing cells had a median mHTT-GFP signal over 25 times higher than control cells (Figure 2D), although the mRNA levels were similar between the two cell lines (Figure 2E). Immunoblot analysis confirmed the higher levels of mHTT in Blos1^s-expressing cells (Figure 2, F–H), although the absolute abundance of mHTT is likely to be underestimated in cells containing aggregated proteins. We solubilized precipitate fractions from cell lysates by passing through a needle; again Blos1^s-expressing cells contained higher levels of mHTT-GFP in the pellet fractions (Figure 2, F and H). Furthermore, depleting control cells of Ire1, but not Xbp1, by RNA interference (RNAi) increased the fraction of transiently transfected mHTT-GFP-expressing cells containing aggregates, whereas mHTT aggregation in Blos1^s-expressing cells was not affected by Ire1 depletion (Figure 2, I–K). Codepletion of both Ire1 and Blos1 restored the baseline level of mHTT aggregation (Figure 2K), suggesting that the primary effect of IRE1 on mHTT aggregation is mediated by degradation of the Blos1 mRNA.

Mutant HTT has previously been shown to be degraded by macroautophagy (MA) (Iwata et al., 2005a,b). Consistently, treatment of our stable cell line that inducibly expresses mHTT-GFP with high levels of doxycycline (4.5 µM) led to higher levels of lipidated LC3B (LC3B-II, Figure 3, A and B), a marker for the induction of MA. However, at lower concentrations of doxycycline (0.9 µM), LC3B-II levels remained low and indistinguishable from those in cells without doxycycline. It was not possible to determine precisely whether the increase in LC3B-II levels in high doxycycline was due to increased induction of MA vs. less efficient degradation of LC3B-II following the fusion of autophagosomes with lysosomes, because inhibiting lysosomal function for the full timescale of the experiment led to extensive cell death. However, including chloroquine, which blocks the acidification of lysosomes, for the final 2 h did lead to increased LC3B-II in all conditions (Figure 3, A and B). To test the idea that MA is responsible for degradation of mHTT only at high expression levels, we used RNAi to deplete the mRNA encoding the essential MA factor ATG7, induced mHTT expression with either 0.9 or 4.5 µM doxycycline, and measured the accumulation of mHTT-GFP by flow cytometry. Under the high-doxycycline conditions, Atg7 knockdown led to a large increase in mHTT levels, whereas under the low-doxycycline conditions, Atg7 knockdown had no effect (Figure 3C). In contrast, knockdown of either Hdac6 (which is important for trafficking aggregating proteins to the MTOC) or Vps22 (which is important for degradation of ubiquitinated aggregates during ER stress; Bae et al., 2019), led to accumulation of mHTT in both doxycycline concentrations (Figure 3C).

These results suggest that mHTT induces MA and is degraded by MA only when expressed at high levels, potentially when larger aggregates begin to form. To further test this idea, we measured both aggregate formation (by microscopy) and MA induction (by immunoblot of LC3B-II) over time in cells expressing Blos1^s, which accumulate aggregates when mHTT expression is induced with either low or high concentrations of doxycycline. The induction of MA coincided with the appearance of large, visible aggregates, after 24 h of 4.5 µM doxycycline or 36 h of 0.9 µM doxycycline (Figure 3, D–F). We therefore propose that while MA is involved in mHTT degradation, a distinct pathway is necessary for its degradation at lower expression levels and/or smaller aggregation states, before large aggregates form. Furthermore, because Blos1^s affects mHTT degradation in both cases, RIDD of Blos1 appears to enhance this MA-independent pathway.

To explore the mechanism of mHTT degradation in the low expression and sub-aggregation state, we depleted cells of various factors by RNAi, induced mHTT expression with 0.9 µM doxycycline, and measured its accumulation by flow cytometry. Similarly to the depletion of Atg7, depletion of Lamp2 (which is important for chaperone-mediated autophagy) did not affect mHTT levels (Figure 3G). However, we observed strong mHTT accumulation in cells depleted of factors involved in the endosomal sorting complex required for transport (ESCRT) pathway, which is responsible for the inward budding of vesicles into LE’s or multivesicular bodies (Henne et al., 2011). The ESCRTs are composed of distinct complexes (ESCRTs -0, -I, -II, -III, and VPS4) that sort cargo and deform the limiting membrane of the endosome to allow for the formation of LEs, which then fuse with lysosomes for cargo degradation. In our experiments, mHTT accumulated following knockdown of the ESCRT-I factor Tsg101, the ESCRT-II factor Vps22, the ESCRT-III accessory factor Alix, and Vps4A, which is essential for the disassembly and recycling of ESCRT machinery (Figure 3, C and G).

ESCRT-0 also appears to be critical for preventing the accumulation of mHTT. ESCRT-0 is a heterotetromer of two HRS and two STAM1 or STAM2 subunits, and it binds both endosomal membranes and ubiquitinated targets and initiates the multivesicular body pathway (Henne et al., 2011). Depletion of Stam1 or 2 led to dramatic accumulation of mHTT (Figure 3G), whereas depletion of Hrs1 led to cell death upon induction of mHTT expression (Figure 3H). Furthermore, depletion of Rab35, which encodes a small GTPase that regulates HRS (Sheehan et al., 2016; Vaz-Silva et al., 2018), also led to strong mHTT accumulation (Figure 3G). Depletion of Stam1, but not Atg7, also led to the accumulation of mHTT aggregates 24 h after transient transfection with low concentrations of the mHTT-GFP plasmid, confirming the effects measured by flow cytometry (Figure S1E).

Based on these data, we conclude that several core components of the ESCRT machinery are important for the degradation of mHTT. ESCRTs are involved in many cellular mechanisms, and directly or indirectly affect all three major autophagic pathways (Vietri et al., 2020), MA, chaperone-mediated autophagy, and endosomal microautophagy (eMI). However, the lack of dependence on MA or chaperone-mediated autophagy for mHTT degradation at low induction levels suggests that the ESCRTs contribute to mHTT degradation directly via eMI (Schuck, 2020). Although the mechanistic details are not yet well understood for mammalian cells, eMI is characterized by the delivery of cytosolic cargo, such as glycolytic proteins, to LEs in a manner that relies on several ESCRT proteins, including TSG101 and VPS4 (Sahu et al., 2011; Vietri et al., 2020).

To further test a potential role for eMI in the degradation of mHTT-GFP, we examined its colocalization with RAB7, an LE marker, in cells transiently transfected under conditions where depletion of Stam1 but not Atg7 affects aggregate accumulation (0.5 µg of plasmid for 24 h, Supplemental Figure S1E). Under these conditions, most cells display bright, diffuse GFP signal, obscuring any potential encapsulation by endocytic organelles. We therefore prepermeabilized cells with digitonin (60 µg/ml, 4°C, 15 min) before fixing and staining for GFP and RAB7. This approach reduced the GFP signal and revealed several detergent-resistant GFP foci in most cells, whereas RAB7 staining appeared similar to that in cells without prepermeabilization. Cells transfected with mHTT-GFP were more likely than untransfected cells to contain enlarged or more prominent/brighter RAB7 foci (Figure 3, I and J), usually one or two per cell. Strikingly, these RAB7 structures colocalized with mHTT-GFP in 71% of the cells (±0.08%, in an average of three independent experiments with 93–96 cells per replicate; Figure 3I). These results are consistent with the idea that mHTT-GFP is encapsulated in LEs during its degradation before forming large juxtanuclear aggregates. As expected due to the transient nature of cargo-containing LEs, these did not account for most of the detergent-resistant GFP foci, which may represent aggregates or other membrane-bound structures, although we did not detect any colocalization with LC3 under these conditions.

We next carried out apoptosis assays to test the relative toxicity of mHTT in both control and Blos1^s-expressing cells after treatment with low or high doxycycline concentrations. Control cells were much more resistant to apoptosis, indicating that degrading Blos1 mRNA not only prevents the accumulation of mHTT aggregates but also protects cells from death. Surprisingly, Blos1^s-expressing cells treated with 0.9 µM dox were positive for both annexin V and propidium iodide, indicative of late apoptosis, whereas cells treated with 4.5 µM dox were more likely to be stained for annexin V only, indicative of early apoptosis (Figure 3K). These differences cannot be explained by the amount of mHTT or the number of aggregates at the time of the assay (36 h after doxycycline addition), as these were indistinguishable between the two doxycycline treatments (Figures 2D and 3D). However, due to the delay in forming large aggregates in the lower expression regime (Figure 3D), we suggest that these cells are exposed to a more toxic subaggregation form of mHTT for a longer time than in the high-expression regime, where mHTT accumulates more rapidly in large aggregates.

To test the generality of these findings and the ability of Blos1 regulation to enhance mHTT degradation in a neuronal cell line, we transfected wt or mHTT-GFP into mouse Neuro2A (N2A) cells, with or without Blos1^s. As in the MC3T3-E1 cells, expression of mHTT led to the degradation of the Blos1 mRNA, and co-expression of Blos1^s led to increased accumulation of mHTT but not wtHTT (Figure 4, A and B). We then asked whether mHTT is degraded by eMI and/or MA in these cells. In control cells not expressing Blos1^s, knockdown of the ESCRT-0 factor Stam1, but not the MA factor Atg7, led to increased mHTT accumulation (Figure 4C). There was no difference between mHTT levels in control vs. Blos1^s cells after depletion of Stam1. In contrast, Atg7 knockdown led to increased mHTT accumulation beyond the effects of Blos1^s expression. Taken together, these data suggest that the effect of Blos1 regulation is primarily on eMI. When this pathway is crippled by Blos1^s expression, knockdown of Stam1 has no further effect, whereas knockdown of Atg7 (which normally does not affect mHTT degradation in these cells) exacerbates Blos1^s expression by preventing mHTT degradation by MA.

Unlike in MC3T3-E1 cells, LAMP1 structures in the N2A cells were predominantly juxtanuclear even in untransfected cells or cells transfected with wtHTT, and this localization was unaffected by Blos1^s expression (Figure 4, D and E). RAB7 structures, in contrast, were peripherally distributed in a higher fraction of cells, and did appear to be more juxtanuclear upon expression of mHTT but not wtHTT (Figure 4, D and F). Blos1^s expression prevented this repositioning, similarly to our results in MC3T3-E1 cells.

Finally, to test whether Blos1 degradation reduces the accumulation of mHTT in primary neurons, we isolated hippocampal neurons from rat E18 embryos and transfected at 15 d in vitro (DIV) with either wt or mHTT-GFP, along with either Blos1^s-Flag or RFP as a control. We then fixed and stained the neurons at DIV 19 using antibodies for GFP and FLAG (Figure 4, G and H). Neurons expressing wtHTT-GFP displayed diffuse, bright GFP signal, whereas neurons expressing mHTT-GFP contained either diffuse signal or large juxtanuclear aggregates. Neurons coexpressing Blos1^s (identified by FLAG staining) were more likely to contain large aggregates than neurons on the same coverslip lacking FLAG signal, or neurons cotransfected with the control RFP (Figure 4, G and H). These data suggest conserved mechanisms of aggregate clearance in cultured cells and neurons.

DISCUSSION

Our overall model is that before forming large, juxtanuclear aggregates, mHTT is degraded by eMI, in a manner that depends on the degradation of Blos1 by the RIDD pathway (Figure 4F). When efficient degradation of mHTT is compromised (as in cells expressing Blos1^s), mHTT accumulates in juxtanuclear aggregates and MA is induced. As indicated by previous work (Ross and Poirier, 2005), we find that large aggregates may be protective, while earlier steps in the aggregation process may be more detrimental to cell survival. We therefore suggest that the encapsulation of smaller oligomers or preaggregated forms of mHTT into LEs is an important mechanism for avoiding their toxicity.

Our conclusions regarding the effects of Blos1 regulation on mHTT aggregate formation are based on cell lines and transiently transfected cells that overexpress Blos1^s, as the endogenous Blos1 gene is intact. Previously we found that this approach in our MC3T3-E1 cell lines does not prevent all clustering of LE/lysosomes at the MTOC, which occurs normally during serum starvation despite overexpression of Blos1^s (Bae et al., 2019). However, other potential consequences of Blos1 overexpression cannot be ruled out, a caveat that we plan to address in the future by more precise genome editing of the Blos1 alleles.

Our previous work suggests that the effect of Blos1 regulation on protein aggregate accumulation is due mainly to disruption of the kinesin adapter complex BORC, rather than the specific loss of the BLOS1 subunit or the disruption of other complexes containing BLOS1 protein, such as BLOC1 (Scott et al., 2018). This is based on the observation that Blos1^s expression can be rescued by depleting cells of LYSPERSIN, a protein subunit of BORC but not BLOC (Bae et al., 2019). This primary role for BORC is consistent with the model that the enhanced degradation of aggregating proteins is achieved by simply moving degradative organelles to the region of the cell where the proteins to be degraded are also sent. On the other hand, we show here that in N2A cells, Blos1 degradation prevents the accumulation of mHTT even when a large fraction of cells constitutively display juxtanuclear LAMP1 and RAB7 structures. This suggests that Blos1 regulation may affect mHTT accumulation in ways unrelated to, or in addition to, the global trafficking and localization of these organelles to the MTOC. In support of this idea, BORC has also been shown to regulate autophagosome-lysosome fusion (Jia et al., 2017) and LE/lysosome size (Yordanov et al., 2019) and may contribute to the degradation of aggregating proteins in other, undiscovered ways.

Regardless of the mechanism, Blos1 regulation appears to enhance the degradation of mHTT through an ESCRT-dependent and MA-independent pathway. Although Blos1 regulation has not been explored in neurodegenerative diseases, the aggregation-prone tau protein, which is involved in the pathogenesis of Alzheimer’s disease, has also been shown to be degraded by both eMI and MA (Caballero et al., 2018; Vaz-Silva et al., 2018). More generally, dysfunction of the ESCRT machinery, including ESCRT-0, has been linked to neurodegeneration and the accumulation of intracellular protein aggregates (Oshima et al., 2016; Kaul and Chakrabarti, 2018).

Why are different degradation pathways used for the clearance of mHTT? It is likely that MA is necessary for the removal of aggregates once they reach a certain size, but why cells would use eMI rather than MA in the case of monomers or smaller oligomers is an open question. We speculate that by enclosing these smaller aggregates within LEs, neurons retain the option of not only degrading these toxic species within their own lysosomes, but also secreting them outside the cell within exosomes, upon fusion of LEs with the plasma membrane. This mechanism, which is emerging as a common feature of neurodegenerative diseases, may help to protect neurons from toxicity when lysosomal capacity is not sufficient, while also potentially explaining the propagation of aggregates to target cells that take up these vesicles (Hill, 2019; Kalluri and LeBleu, 2020).

MATERIALS AND METHODS

Request a protocol through Bio-protocol.

Supplementary Material