Lou Gehrig’s Disease (ALS): UBQLN2 Mutations Strike Out of Phase

Abstract

In this issue of Structure, Dao et al. (2019) report that ALS-linked mutations in the Pxx domain of Ubiquilin 2 (UBQLN2) differentially influence the protein’s phase separation abilities. The affect is by reducing the temperature and UBQLN2 concentration necessary for liquid-liquid phase separation droplet formation and by modulating UBQLN2 oligomerization.

Amyotrophic lateral sclerosis (ALS), also called Lou Gehrig’s disease after the famous New York Yankees baseball player who succumbed to the disease, is a rapid progressive fatal neurodegenerative disease causing loss of upper and lower motor neurons. Although the cause of the vast majority of ALS cases is not known, approximately 10% of cases are linked to mutations in at least 30 genes. In 2011, five different missense mutations were identified in five unrelated families in Ubiquilin 2 (UBQLN2), located on the X chromosome, by linkage to ALS or ALS with frontotemporal dementia (ALS/FTD) (Deng et al., 2011). ALS is typically associated with loss of voluntary muscle movement, whereas FTD is associated with loss of cognitive function. However, it is estimated that 50% of ALS patients develop cognitive impairments. The five mutations all involved changes in proline residues (P497H, P497S, P506T, P509S, and P525S) in a highly unusual Pxx repeat that is found in UBQLN2 but not in any other UBQLN family member. Since then, many additional mutations in or surrounding the Pxx region have been identified, suggesting a crucial role of the region in causing disease. However, how mutations in this region cause disease is unclear. In this issue of Structure, Dao et al. (2019) report on the biophysical properties of the Pxx-containing portion of the UBQLN2 polypeptide, showing that ALS mutations in the region alter properties of the protein to phase separate, which may have relevance in disease.

The Pxx motif is found within the central domain of the 624-amino acid UBQLN2 polypeptide, which possesses, like other UBQLN proteins, several structural features indicative of a role in protein quality control. The human genome contains four UBQLN proteins encoded by separate genes. The proteins are all ~600 amino acids long and are characterized by short highly conserved UBL and UBA domains at their N- and C-termini, respectively, and a longer, more variable central domain. A key function of UBQLN proteins is to shuttle proteins to the proteasome for degradation by interaction of the UBA domain with ubiquitin chains conjugated onto target proteins and interaction of the UBL domain with subunits of the proteasome (Chang and Monteiro, 2015). UBQLN proteins also contain variable numbers of STI1-like motifs located in the central domain that bind heat shock proteins, consistent with evidence they support chaperone activity. Additionally, the proteins function to clear cytoplasmic material through the autophagy-lysosomal pathway (Şentürk et al., 2019). Inactivation of rodent UBQLN2 results in no overt phenotype, suggesting that ALS/FTD mutations in UBQLN2 are unlikely to result from simple loss of protein function. Instead, studies in cells and animals have revealed that these mutations disturb proteostasis by interfering with proteasomal degradation of proteins and by impeding autophagy (Chang and Monteiro, 2015; Chen et al., 2018). Animal models that recapitulate the human disease have been developed (Gorrie et al., 2014; Le et al., 2016; Chen et al., 2018). A remarkable feature seen in the models is the accumulation of UBQLN2 inclusions in a highly distinctive and unusual pattern in neuronal tissue similar to that found in humans, suggesting a propensity of the mutant proteins to aggregate (Gorrie et al., 2014; Le et al., 2016; Sharkey et al., 2018).

Castaneda and colleagues (Dao et al., 2019) focused on the peculiar property of UBQLN2 to undergo liquid-liquid phase separation (LLPS). They build on previous studies (Dao et al., 2018) showing that UBQLN2 can undergo LLPS under physiologic conditions and that the Pxx region plays an important role in modulating this behavior. Changes in LLPS behavior regulate the dynamics of assembly and dissolution of membraneless organelles that play important functions in cells. One such organelle is stress granules (SGs), which form in response to biotic stress (Dao et al., 2018). SGs dissipate once the stress is dispelled. However, persistence of SGs can drive conversion of proteins within the granules to more gel-like forms and stable aggregates. Interestingly, several proteins involved in ALS localize to SGs, including UBQLN2 (Dao et al., 2018; Alexander et al., 2018; Sharkey et al., 2018). Insights into how ALS mutations in UBQLN2 protein affect its LLPS behavior could therefore be important.

In the present report, Castaneda and colleagues (Dao et al., 2019) used a combination of biophysical, structural, and microscopic approaches to study 11 different ALS mutations in UBQLN2, focusing principally on how they affect LLPS and protein oligomerization. The 11 mutants selected (T487I, A488T, P497H, P497L, P497S, P500S, P506A, P506S, P506T, P509S, and P525S) are all in the Pxx-containing region of UBQLN2. The mean age of onset for each of the mutants in males and females and their disease duration is shown in Figure 1. A few trends are apparent. Males have earlier age of onset than females for the same mutation, which is probably due to differences in UBQLN2 gene dosage from inherent differences in having XY and XX chromosomes. Additionally, mutations that trigger earlier onset of disease seem more aggressive, causing both ALS and FTD, whereas the mutations that have later onset seem to cause mainly ALS.

Figure 1.

Onset and Disease Progression for ALS or FTD Mean age of onset of ALS or FTD and disease progression for the different mutations that were studied (Dao et al., 2019). The mutants that showed increased oligomerization in the study are indicated with a “+” symbol.

Dao et al. (2018) used the C-terminal 450–624 amino acid portion of UBQLN2 for all of their investigations, as this region was found to be sufficient for LLPS and localization to stress-induced puncta. The fragment contains a portion of the STI1-II motif (residues 450–486), the Pxx region (487–538), a low-complexity domain (339–462), and the UBA domain (577–620). They monitored LLPS formation and dissolution by spectrophoto-metric changes in turbidity and by microscopy, at different temperatures and salt conditions, obtaining similar results. The mutants were classified into two groups. One group, composed of A488T, P500S, P509S, and P525S mutants, all exhibited WT-like reversible LLPS characteristics, requiring similar temperature to phase separate, and they formed round droplets that grew in size over time by fusion with one another. The second group, composed of T487I, P497H, P497L, P497S, P506A, P506S, and P506T mutants, phase separated at lower temperatures into abnormal amorphous droplets or aggregates. Next, they examined whether changes in oligomerization of the proteins could explain the difference. Indeed, using size exclusion chromatography and nuclear magnetic resonance (NMR) spectroscopy, they found the mutants in the latter group self-associated more than the former group, forming increasingly higher-order oligomeric species. However, they found that the mutants caused only subtle changes in the UBQLN2 protein structure.

Previously, Castaneda and colleagues (Dao et al., 2018) found that addition of ubiquitin could dissolve phase droplets formed by WT UBQLN2 protein and that this was dependent on ubiquitin binding to the UBA domain of UBQLN2. Therefore, they tested whether addition of ubiquitin affected LLPS formed by the mutants and found that it was capable of dissolving all of them, including protein aggregates formed by the T487I and P497S mutants. Further studies revealed that ubiquitin addition did not alter the ability of the mutants to oligomerize. Instead they found that binding of ubiquitin disrupts intra-molecular interactions between the UBA domain and the STI1 and PXX regions in the UBQLN2 polypeptide. Based on these results, the authors hypothesized that the UBA and Pxx regions of UBQLN2 may transiently interact to modulate UBQLN2 LLPS and that ubiquitin binding to the UBA domain interferes with this interaction.

Taken together, the observations made throughout this study support a model in which Pxx mutations that cause increased UBQLN2 oligomerization lead to more significant alterations in LLPS behavior and material properties of UBQLN2 droplets, with the most extreme cases resulting in protein aggregation. Interestingly, the effect of Pxx mutations on UBQLN2 LLPS behavior and oligomerization was highly dependent on the type and position of the mutation. Importantly, mutations to hydrophobic residues (e.g., T487I, P497L, and P506A) increased oligomerization, which suggests that UBQLN2 LLPS may be driven by hydrophobicity. Ultimately, the observation that some ALS-linked mutations alter oligomerization and LLPS behavior while others do not suggest that mutations to the Pxx region of UBQLN2 may lead to disease pathogenesis through different mechanisms.

These in vitro studies of the underlying mechanism by which mutations perturb LLPS formation and oligomerization and lead to aggregation of UBQLN2 proteins provide a framework to understand ALS pathogenesis. However, it is important to relate these findings to the function and dysfunction of the proteins in vivo. In addition, it is essential to know whether these and other ALS mutations in UBQLN2 perturb LLPS behavior using full-length UBQLN2 protein, whose folding could be different than the truncated version used in the present study. It is also important to know if the phenomena observed are related to the high concentrations of protein used for the assays, which were 10–500 times the level of UBQLN2 found in cells (Dao et al., 2018). Many proteins, including other UBQLN isoforms, heat shock proteins, ER-associated degradation factors, and hnRNP proteins, have been shown to interact with the central domain of UBQLN2, which is the same region implicated in LLPS. How ALS mutations influence these interactions needs to be understood in terms of how they affect LLPS and the physiologic consequences of this behavior. An important consideration is the role that ubiquitin binding, including the potential spatiotemporal effects of different types of ubiquitin chains, has in regulating LLPS and UBQLN2 function in vivo. It is worth noting that most of the ALS mutants that were least disruptive in LLPS behavior cause later age of onset of ALS, hinting that the phenomenon may be somehow tied to pathogenesis. Another fascinating issue that still needs to be understood is why males develop disease earlier than females with the same mutations, and whether age of onset is related to UBQLN2 gene dosage and X chromosome inactivation.