Lipid perturbation compromises UPR-mediated ER homeostasis as a result of premature degradation of membrane proteins

Phospholipid homeostasis in biological membranes is essential to maintain cellular functions of organelles such as the endoplasmic reticulum. Phospholipid perturbation has been associated to non-alcoholic fatty liver disease, obesity and other metabolic disorders. However, in most cases, the biological significance of lipid disequilibrium remains unclear. Previously, we reported that Saccharomyces cerevisiae adapts to lipid disequilibrium by upregulating several protein quality control pathways such as the endoplasmic reticulum-associated degradation (ERAD) pathway and the unfolded protein response. Surprisingly, we observed certain ER-resident transmembrane proteins (TPs), part of the UPR programme, to be destabilised under lipid perturbation (LP). Among these, Sbh1 was prematurely degraded by dissociating from the Sec61 complex due to fatty acid remodelling and membrane stiffening of the ER. Sbh1 is targeted for degradation through its highly conserved lysine residue near the membrane in a Doa10-dependent manner. Premature removal of key ER-resident TPs might be an underlying cause of chronic ER stress in metabolic disorders.


74
Phospholipid homeostasis is crucial in the maintenance of various cellular processes and functions. a role in the translocation of proteins into the ER (Fig. S1). It is non-essential for translocation but can 165 cause a translocation defect when both Sbh1 and its homolog, Sbh2, are absent (Feng et al., 2007).

167
To assess the stability of TP candidates during lipid imbalance, cycloheximide chase assay was 168 performed in WT and opi3Δ strains. Time points were taken from adding translation inhibitor 169 cycloheximide to measure the half-life of HA tagged-proteins over a period of one hour. Emc4, Nsg2, 170 and Sbh1 half-life were found to be significantly reduced under lipid disequilibrium (Fig. 1 C). No 171 significant decrease in Cue1-HA protein level was detected in opi3Δ (P value of 0.051). After one hour 172 chase, protein levels of Emc4, Nsg2, and Sbh1 were found to be 27%, 41%, and 58% lower in opi3Δ, 173 respectively, compared to WT. This suggests that the UPR programme transcriptionally upregulates 174 genes to restore ER homeostasis upon LP while a subset of TPs is degraded prematurely. To ensure that Cue1, Emc4, Nsg2, and Sbh1 are not prematurely degraded due to mislocalisation, 179 indirect immunofluorescence was performed with the four protein candidates. They were found to co-180 localise with the molecular chaperone ER marker Kar2 in opi3Δ as in WT (Fig. 2 A). To further confirm 181 the four TPs remain inserted into the ER membrane, cell extracts from WT and opi3Δ strains expressing 182 the four candidates were treated with alkali to strip soluble and peripherally associated protein from 183 membranes (Stukey, McDonough, & Martin, 1989). Membranes and alkali-extracted proteins were 184 separated by high speed centrifugation and detected by immunoblots. The four candidates were 185 fractionated exclusively within the membrane pellet (Fig. 2 B). Together, these results suggest that 186 localisation and integration of the four TPs into the ER membrane were unaffected by 187 phosphatidylcholine depletion. To ensure the premature degradation of these four proteins is not due 188 to inverted topology, we performed proteinase K (PK) digestion from isolated microsomes (Fig. 2 C). In

189
WT cells, C-termini HA tags of Cue1-HA, Emc4-HA and Nsg2-HA are found in the cytosol. Thus, HA 190 tag will be cleaved off if the topology is intact while a peptide will be detected if the topology is inverted 191 upon PK digestion. The three proteins were found to be fully digested under LP and the predicted 192 smaller protein fragments of 23.7, 8.53, and 5.8 kDa were not detected for Cue1-HA, Emc4-HA, and HA tag is found in the ER lumen, hence the predicted protein fragment of 10.5 kDa was detected in 195 both WT and opi3Δ strains. Typically, tail-anchored proteins are tagged at the N-termini as C-termini 196 interact with the Get complex for insertion into the ER membrane (F. Wang, Whynot, Tung, & Denic, 197 2011). Thus, this result combined with alkaline carbonate extraction show that adding a C-terminus HA 198 tag to Sbh1 did not interfere with its integration into the ER membrane. Together, these results suggest 199 that the four TP topologies are not affected from LP and thus eliminating this factor as an underlying

206
and Sbh1-HA to the level found in WT (Fig. 2 D). Subsequently, we concentrated our effort on Sbh1 to 207 better understand how a membrane protein is destabilised from LP.

209
The UPR is strongly activated in response to LP (

219
To characterise lipid remodelling specific to the ER in opi3Δ that might contribute to the premature 220 degradation of TPs, we analysed the fatty acid (FA) composition of whole cells and fractionated microsomes from WT and opi3Δ. Overall, total FAs in opi3Δ cells increased by 96% when compared to 222 WT (Fig. 3 B). This was expected as opi3Δ accumulates large lipid droplets (Fei et al., 2011; Thibault saturated FAs is more drastic in opi3Δ microsomes compared to whole cells (Fig. 3 B). In addition of        To further validate that Sbh1 is degraded in a Doa10-dependent manner, we carried out cycloheximide 311 chase assay to monitor Sbh1 stability in different ERAD mutants. Sbh1 was found to be fully stabilised Sbh1 degradation is not affected by genetically inhibiting vacuolar protein degradation using pep4Δ but 317 showed dependency to Cue1, a conserved element in both the Doa10 and Hrd1 complexes (Fig. S6).

318
Together with the MYTH data, it suggests that Sbh1 is exclusively targeted for degradation by the ERAD 319 Doa10 complex through a cytosolic lesion.    mono-methylated, it has physical properties similar to PE (Fig. 1 A). The increase in membrane 385 curvature from the replacement of PC to MMPE may induce cells to decrease their fatty acid chain 386 lengths in accordance to the seminal Helfrich theory of membrane bending elasticity (Fig. 3 A; cell suggests either the ER is more susceptible to LP due to the minimal presence of ergosterol at the 389 ER (Zinser et al., 1993) or cells respond more aggressively to the ER membrane bilayer disruption to 390 alleviate ER stress. Accordingly, a rise in membrane lipid packing from elevated saturated fatty acids 391 will reduce the potency to form curvatures (Fig. 3 B).

393
The remodelling of the ER to alleviate negative membrane curvature stress, induced from high PE and  (Fig. 4, A and B). The loss of Sbh1 405 interaction to the Sec61 complex supports its susceptibility to be prematurely degraded as its assembly

408
However, Sbh1 is not required for the stability of Sec61 in complex with Sss1 (Fig. 4 C). In contrast,

409
Sbh1 was found to interact with Sec61 by cross-linked co-IP. This might be due to new electrostatic or  reported for its homolog Sbh2 (Habeck et al., 2015). Thus, a decrease of PC clearly targets Sbh1 for 425 degradation from a change in the biophysical property of the membrane (Fig. 5 E). It remains to be 426 determined if Sbh1 degradation mechanism from LP applies to other unstable TPs identified (Fig. 1 B).

433
Despite this robust stress response under LP, the UPR programme fails to increase the expression 434 level of a subset of TPs (Fig. 1 C). The premature degradation of these TPs can prevent an effective 435 proteostatic response especially under prolonged LP (Fig. 6). ER stress induced from a temporary lipid 436 perturbation will result in the upregulation of UPR target genes and consequently ER homeostasis.

437
However, in the context of fatty liver, prolonged LP might prevent cells to reach ER homeostasis from 438 the premature degradation of some key UPR target TPs. Therefore, this will lead to chronic ER stress

475
Plasmids and primers used in this study are listed in Table S3 and S4, respectively. Plasmids were 476 constructed using standard cloning protocols. All coding sequences of constructs used in this study 477 were sequenced in their entirety. The plasmid pGT0179, pGT0181, pGT0183, and pGT0185, were

508
Indirect immunofluorescence was carried out as previously described (Spear & Ng, 2003). Typically, 509 cells were grown to early log phase at 30°C in selective synthetic complete media, fixed in 10% 510 formaldehyde and permeabilised. After blocking with 3% BSA, staining was performed using anti-HA

1015
The degradation of Sbh1-HA was analysed in WT and opi3Δ cells containing control vector (ve) or Figure S3. Lack of OPI3 leads to broad remodelling of fatty acids.

1021
Percentage of fatty acid chain lengths to total fatty acids in opi3Δ is compared to WT. Total fatty acid 1022 content in microsomes of WT and opi3Δ were quantified by gas chromatography after FAME

1049
Cells were grown to early log phase at 30˚C in selective synthetic complete media. UPR induction was 1050 measured using a UPRE-LacZ reporter assay. Tm, tunicamycin.