SEL1L-HRD1 interaction is required to form a functional HRD1 ERAD complex

The SEL1L-HRD1 protein complex represents the most conserved branch of endoplasmic reticulum (ER)-associated degradation (ERAD). Despite recent advances in both mouse models and humans, in vivo evidence for the importance of SEL1L in the ERAD complex formation and its (patho-)physiological relevance in mammals remains limited. Here we report that SEL1L variant p.Ser658Pro (SEL1LS658P) is a pathogenic hypomorphic mutation, causing partial embryonic lethality, developmental delay, and early-onset cerebellar ataxia in homozygous mice carrying the bi-allelic variant. Biochemical analyses reveal that SEL1LS658P variant not only reduces the protein stability of SEL1L, but attenuates the SEL1L-HRD1 interaction, likely via electrostatic repulsion between SEL1L F668 and HRD1 Y30 residues. Proteomic screens of SEL1L and HRD1 interactomes reveal that SEL1L-HRD1 interaction is a prerequisite for the formation of a functional HRD1 ERAD complex, as SEL1L is required for the recruitment of E2 enzyme UBE2J1 as well as DERLIN to HRD1. These data not only establish the disease relevance of SEL1L-HRD1 ERAD, but also provide additional insight into the formation of a functional HRD1 ERAD complex.

7) "Hence, these data show that KI mice exhibit microcephaly, with no massive cell death or inflammafion in the brain."It should be noted that the effect is mild.

8) "
In line with our previous study that UPR sensor IRE1α is an ERAD substrate 23, its protein level was increased by ~4 folds; however, neither IRE1α phosphorylafion nor splicing of its downstream effector Xbp1 mRNA was elevated…" There should be a comment on this in the results or discussion.How can it be explained that there is no increased XBP1 splicing with such an increase in IRE1? 9) Fig. 3b, why is total PERK much increased?Comment on this.10) "…which was aftenuated upon the over-expression of SEL1L WT (lane 3-5 vs. 2), but not SEL1L S658P (lane 6-8 vs. 3-5, Fig. 4a and quanfitated in Fig. 4b)."There is sfill some effect , perhaps "but much less SEL1L S658P".11) Secfion "Sequence and structural analyses of SEL1LS658P variant" analyzes the predicfions on S658P proximity to HRD1 loops but does not comment on the proximity to OS-9.In light of this the results that show no effect of the mutafion on SEL1L interacfion with OS-9 might be unexpected, so the proximity to OS-9 should be menfioned in this secfion and the lack of interference commented in the Discussion.
12) Fig. 7d: It seems that the names Group I and Group II were swapped here.
13) "…the upregulafion of ER chaperones to increase folding efficiency, enhanced aggregafion and sequestrafion of misfolded proteins to hence aftenuate proteotoxicity and the acfivafion of ER-phagy to clear misfolded protein aggregates in the ER 62,63."It could be added here that perhaps in vivo there is compensafion by ERAD involving alternafive E3 ligases.

Reviewer #3 (Remarks to the Author):
In this manuscript, the authors explore the consequences of a disease-associated mutafion in SEL1L, recapitulafing its effects in vivo to analyse phenotypic effects and subsequently delve into the molecular mechanics of the observed effects in cellular models.They highlight intriguing features of the transgenic animals caused by the mutafion and link them to the defect in ERAD, as demonstrated in their cellular model.The authors underscore the potenfial selecfive vulnerability of Purkinje cells to ERAD disrupfion and the developmental nature of the relevant pathology.The adept use of predicfive structure analysis via AlphaFold has enabled the authors to deepen their understanding of the consequences of SEL1L mutafion, which they further validated through co-IP interacfion analysis upon SEL1 mutagenesis.The manuscript also characterizes the relevant ER stress status associated with SEL1L defect.Integrafing their results, the authors suggest a potenfial selecfive vulnerability of Purkinje cells to ERAD perturbafion, hypothesising the existence of a specific ERAD substrate that exposes these cells to SEL1 malfuncfion.
Overall, the conclusions appear to be drawn from comprehensive, high-quality experimental data and represent a significant advancement in understanding the protein's role in ERAD and the implicafions of its malfuncfion for brain health.This informafion will be valuable for the field and broadly as it highlights and mechanisfically describe the connecfion between neuronal health and ERAD invifing further elucidafion of this process.
A few points the authors should consider: • In Fig. 4, the effect of the SEL1L mutafion on the degradafion rate of different substrates in the cycloheximide chase appears to be parficularly prominent for CD147 as an ERAD substrate.The differences for various substrates should be contextualised with their overall half-lives.Given this substrate's sensifivity to SEL1L acfivity, experiments showing the compensatory effect of the SEL1L S658P/F668Y variant should be presented.This would substanfiate the conclusions drawn from the observed parfial reversal of HMW accumulafion, which isn't a direct ERAD-reporfing observafion.
• Supported by the structural analysis, in Fig. 6, the authors invesfigate how SEL1LS658P aftenuates the SEL1L-HRD1 interacfion and aftribute this to the interface between F668 (SEL1L) and Y30 (HRD1).While the results back their hypothesis, addifional evidence would solidify this conclusion.Demonstrafing that mutafing HRD1, targefing Y30, produces a similar effect would be pivotal.As the co-IP method doesn't conclusively indicate the directness of interacfions, these further evidences would be needed for establishing the effect on SEL1-HRD1 interacfion.Affinity measurements between the variants of the two proteins would also provide a more definifive support for this core finding.
• In Extended Data Fig. 6c, the authors suggest that the asparagine mutant (F668N) doesn't disrupt the interacfion between SEL1L and HRD1.Given that asparagine isn't an aromafic residue, the authors should discuss these results in the context of their aromafic-aromafic interacfion theory.
• The authors should consider elaborafing on the rafionale behind defining the phenotype as ataxia.
Minor Correcfions: • Page 7: Change "rafion" to "rafio?" • Page 9: Correct "whether" and "rescued." • Should the F668 mutant that counteracts the perturbafion by the S658P mutafion be "F668Y" instead of "F668W?" • In Fig. 7d, should the group labels be switched?Should the purple be Group I, and the blue be Group II?
• In Fig. 7c, the heatmap for DERL2 in the SEL1L IP panel seems inconsistent with other data.The authors should address the potenfial reasons for DERL2's absence in mass spectrometry.
Reviewer #4 (Remarks to the Author): I understand the authors' efforts in trying to show the physiological importance of interacfion between Sel1L and HRD1 by using the pathological mutafion SEL1L(S658P) idenfified in dog suffering cerebellar ataxia.They generated SEL1L(S658P) knock-in (KI) mice and analyzed the cerebellar ataxia phenotype of homozygous KI mice in detail.The importance of the SEL1L(S658P) mutafion was evident from genefic analysis in dog lines, the data obtained from the SEL1L(S658P) cock-in mice is only confirmatory.On the other hand, they showed that SEL1L(S658P) mutafion reduced the interacfion with HRD1 in HEK293 cells and hypothesized that the aftenuafion of SEL1L-HRD1 interacfion causes HRD1 dysfuncfion by creafing a repulsion between SEL1L(F668) and HRD1(Y30) residues in silico.However, there is no physiological relevance between the data obtained from mice experiments and those from HEK293 cells or in silico.Their claim that the SEL1L(S658P) mutafion causes only marginal ER stress and does not acfivate the caspase pathway.However, it is not clear whether ERAD is inhibited or HRD1 funcfion is impaired in the cerebellum of the SEL1L(S658P) KI mouse.The two topics of the manuscript (the interacfion of Sel1L with HRD1 in HEK293 cells and the pathological mechanisms of the SEL1L(S658P) KI mouse phenotypes) are not well integrated, and the manuscript would be befter presented as two separate papers.

Other points
It is unclear why a half of SEL1L(S658P) homozygous KI mice are lethal.
In the beam walking test, quanfitafive data on the number of fimes foot slipped is needed.
P6.Not shown data should be shown.
Since CHOP is considerably elevated in the cerebellum of Finnish Hounds lines (ref 47), this should be carefully examined in KI mice as well.
They claimed that neurons including Purkinje cells adapt to the expression of SEL1L(S658P) without elicifing an overt ER stress or cell death (P7).However, because the cerebellum contains many Bergmann glia and other cells, more detailed studies must be carefully conducted to confirm Purkinje cell-specific changes.
We thank all three reviewers for their insightful and constructive comments.We now have carefully addressed all the comments from the reviewers, which have been instrumental to further improve our manuscript.

Reviewer #1 (Remarks to the Author):
This is an interesting study on the function of SEL1L-HRD1 interaction in ERAD and the impairment in mice carrying a variant SEL1LS658P.It includes extensive well devised experimental data.The following concerns should be addressed: There are no page or line numbers, which would facilitate the review.I will copy fragments of the article when necessary: We have now added the page and line numbers in the manuscript.
1) "These data suggest that SEL1L S658P causes a physical collision between SEL1L F668 and HRD1 Y30, thereby attenuating SEL1L-HRD1 interaction."A major concern is that to have a definitive proof of this, mutations in HRD1 Y30 should be made and tested.Alternatively, the conclusions should be moderated (here and in the Abstract and Discussion), speculating that the results are consistent with this.
We thank the reviewer for this great comment.As suggested, we have now generated HRD1 Y30A, Y30D, Y30K and Y30F mutants and assessed their impact on HRD1 interaction with SEL1L and other ERAD components.Our IP data showed that HRD1-Y30 mutated to A, D or K significantly disrupted the SEL1L-HRD1 interaction by approximately 80-90%, while Y30F resulted in a 30% reduction (Response Figure 1).The disruption of SEL1L-HRD1 interaction abolished the interaction of HRD1 with the ER lectin OS9, while having no impact on the interaction with FAM8A1 (Response Figure 1).This data suggests that the HRD1 Y30 is critical determinant for the SEL1L-HRD1 interaction and complex formation.In addition, we have tone down the conclusion throughout the text.This data is now shown in Extended Data Fig. 7c in the revised manuscript.

Response Figure 1. HRD1 Y30 is critical for the SEL1L-HRD1 interaction.
Immunoprecipitation of FLAG-agarose in HRD1 -/-HEK293T cells transfected with indicated HRD1-FLAG constructs to exam the interaction with SEL1L, OS9 and FAM8A1, with quantitation shown below the gel as means from two independent repeats.
Result: "We first examine whether SEL1L F668 or HRD1 Y30 is critical for the SEL1L-HRD1 interaction.Indeed, HRD1 Y30 mutated to Ala (A), Asp (D) or (K) significantly disrupted the SEL1L-HRD1 interaction by 80-90%, and 30% when mutated to Phe (F).The disruption of SEL1L-HRD1 interaction abolished the interaction of HRD1 with OS9, while having no impact on the interaction with FAM8A1." Result: "These data suggested that SEL1L S658P may cause a physical collision between SEL1L and HRD1 via SEL1L F668 and HRD1 Y30 residues, while having no effect on SEL1L-OS9/ERLEC1 interaction.However, the definitive support for this model will require the affinity measurements between the two proteins." 2) Abstract "definitive evidence for the importance of SEL1L in HRD1 ERAD is lacking."This statement is exaggerated as there are many previous studies on this topic.We thank the reviewer for this great comment.We now have changed this statement in the Abstract.
Abstract: "in vivo evidence for the importance of SEL1L in the ERAD complex formation and its (patho-)physiological relevance in mammals remains limited." 3) Abstract "…causes HRD1 dysfunction by generating electrostatic repulsion between SEL1L F668 and HRD1 Y30 residues."This is not definitely proven, as I comment above, it should be moderated.See response to Point 1 and Response Figure 1.

4)
Abstract "…but the E2 UBE2J1 and retrotranslocon DERLIN, to HRD1."The results show Derlin2, which in any case has not been formally shown to be a part of a retrotranslocon, so perhaps "putative retrotranslocon component".This correction should also be introduced in other parts of the text when the protein is mentioned.
Introduction: "…subsequently exacted from the ER membrane by the cytosolic AAA-ATPase Cdc48/VCP for proteasomal degradation.Human HERP (Usa1p in yeast) promotes HRD1 oligomerization as well as the formation of ERAD complex in vitro, which is required for substrate retrotranslocation." 6) "These data show that SEL1LS658P KI mice exhibit growth retardation, and signs of early onset nonprogressive ataxia, establishing the disease-causality and pathogenicity of this allele."It should be noted that the effects are mild.We thank the reviewer for this comment.We have now revised it in the text and pasted below.
Result: "These data show that SEL1L S658P KI mice exhibit mild growth retardation, and signs of early onset non-progressive mild ataxia, establishing the disease-causality and pathogenicity of this allele." 7) "Hence, these data show that KI mice exhibit microcephaly, with no massive cell death or inflammation in the brain."It should be noted that the effect is mild.We thank the reviewer for this comment.We have now revised it on Page 6 and pasted below.
Result: "Hence, these data show that KI mice exhibit mild microcephaly, with no massive cell death or inflammation in the brain."

8) "
In line with our previous study that UPR sensor IRE1α is an ERAD substrate 23, its protein level was increased by ~4 folds; however, neither IRE1α phosphorylation nor splicing of its downstream effector Xbp1 mRNA was elevated…" There should be a comment on this in the results or discussion.How can it be explained that there is no increased XBP1 splicing with such an increase in IRE1?
We thank the reviewer for this great comment and apologize for the confusion.The IRE1α protein levels in the KI cerebellum were increased by ~4 folds (Response Figure 2a), in line with the notion that UPR sensor IRE1α is an HRD1-SEL1L ERAD substrate and ERAD deficiency causes IRE1α protein stabilization and accumulation 1 .However, IRE1α phosphorylation was not elevated as shown in Response Figure 2a and consistently, its downstream effector Xbp1 mRNA splicing was not significantly increased in the KI cerebellum (Response Figure 2b).We now have added a positive control (Liver+TM) for Xbp1 mRNA splicing as shown in Response Figure 2b.This data suggested that the SEL1L S658P KI cerebellum does not instigate an overt IRE1α activation.Response Figure 2 is now shown in Fig. 4a and  4d.We have now commented it in the Discussion and pasted below.The most likely explanation for no increased Xbp1 splicing or UPR activation is cellular adaptation.We propose that there is an activation of adaptive mechanisms in response to the expression ERAD variants, such as the upregulation of ER chaperones GRP94, PDI and BiP to increase folding efficiency (Response Figure 2c) and/or ER-phagy to degrade misfolded protein aggregates in the absence of SEL1L-HRD1 ERAD as we recently shown in ERAD KO cells/tissues 2,3 .

Discussion: "The lack of an overt UPR in the KI mice is likely due to various adaptive mechanisms in response to a hypomorphic variant, including, but not limited to, the upregulation of ER chaperones to increase folding efficiency, enhanced aggregation and sequestration of misfolded proteins to hence attenuate proteotoxicity, the activation of ER-phagy to clear protein aggregates, and/or ERAD involving alternative E3 ligase to clear misfolded proteins in the ER."
9) Fig. 3b, why is total PERK much increased?Comment on this.We thank the reviewer for this great comment.The reason underlying the increased PERK in the KI cerebellum is currently unclear, although it is likely caused by the elevated PERK gene transcription.
Results: "In line with these findings, turnover of a known ERAD substrate, a disease mutant of pro-arginine vasopressin (proAVP) at residue 57 (Gly-to-Ser, Gly57Ser), was attenuated in SEL1L S658P transfected SEL1L -/- HEK293T cells, leading to its accumulation and the formation of HMW aggregates (lane 6-8 vs. 3-5, Fig. 5h and quantitated in Fig. 5i)." 11) Section "Sequence and structural analyses of SEL1LS658P variant" analyzes the predictions on S658P proximity to HRD1 loops but does not comment on the proximity to OS-9.In light of this the results that show no effect of the mutation on SEL1L interaction with OS-9 might be unexpected, so the proximity to OS-9 should be mentioned in this section and the lack of interference commented in the Discussion.We thank the reviewer for this great comment.We have now added a side view of the SEL1L-OS9 interface around the SEL1L 658 position.Indeed, as suggested by the reviewer, SEL1L E659 and R655 may interact with OS9 (Response Figure 3c).However, further immunoprecipitation and biochemical data showed that SEL1L S658P mutation does not affect its interaction with OS9 and ERLEC1 (Fig. 6e) for the unknown reason.
We have made changes accordingly in the revised manuscript to reflect these findings.
Response Figure 3 is now shown in Fig. 6cd and Extended Data Fig. 6c in the revised manuscript.We also added a discussion on this point in the Results.Results: "On the other hand, SEL1L S658 is in proximity to residues E659 and R655 of SEL1L, which may be involved in the interaction with OS9."

Response
Result: "These data suggested that SEL1L S658P may cause a physical collision between SEL1L and HRD1 via SEL1L F668 and HRD1 Y30 residues, while having no effect on SEL1L-OS9/ERLEC1 interaction.However, the definitive support for this model will require the affinity measurements between the two proteins."12) Fig. 7d: It seems that the names Group I and Group II were swapped here.We thank the reviewer for this comment and have now corrected Group I and II in the Figure .13) "…the upregulation of ER chaperones to increase folding efficiency, enhanced aggregation and sequestration of misfolded proteins to hence attenuate proteotoxicity and the activation of ER-phagy to clear misfolded protein aggregates in the ER 62,63."It could be added here that perhaps in vivo there is compensation by ERAD involving alternative E3 ligases.We thank the reviewer for this great comment and have included the Discussion on the alternative E3 ligases and pasted blow.
Discussion: "The lack of an overt UPR in the KI mice is likely due to various adaptive mechanisms in response to a hypomorphic variant, including, but not limited to, the upregulation of ER chaperones to increase folding efficiency, enhanced aggregation and sequestration of misfolded proteins to hence attenuate proteotoxicity, the activation of ER-phagy to clear protein aggregates, and/or ERAD involving alternative E3 ligase to clear misfolded proteins in the ER." 14) The English is good in general, but I spotted several typos, such as "signification subfraction ", "Structral predication ", "whther the resuced interacion".The text should be carefully edited.We thank the reviewer for pointing out those typos.We have now carefully edited the text and corrected typos.

Reviewer #3 (Remarks to the Author):
In this manuscript, the authors explore the consequences of a disease-associated mutation in SEL1L, recapitulating its effects in vivo to analyse phenotypic effects and subsequently delve into the molecular mechanics of the observed effects in cellular models.They highlight intriguing features of the transgenic animals caused by the mutation and link them to the defect in ERAD, as demonstrated in their cellular model.The authors underscore the potential selective vulnerability of Purkinje cells to ERAD disruption and the developmental nature of the relevant pathology.The adept use of predictive structure analysis via AlphaFold has enabled the authors to deepen their understanding of the consequences of SEL1L mutation, which they further validated through co-IP interaction analysis upon SEL1 mutagenesis.The manuscript also characterizes the relevant ER stress status associated with SEL1L defect.Integrating their results, the authors suggest a potential selective vulnerability of Purkinje cells to ERAD perturbation, hypothesising the existence of a specific ERAD substrate that exposes these cells to SEL1 malfunction.Overall, the conclusions appear to be drawn from comprehensive, high-quality experimental data and represent a significant advancement in understanding the protein's role in ERAD and the implications of its malfunction for brain health.This information will be valuable for the field and broadly as it highlights and mechanistically describe the connection between neuronal health and ERAD inviting further elucidation of this process.We thank this reviewer for the constructive comments.

A few points the authors should consider:
• In Fig. 4, the effect of the SEL1L mutation on the degradation rate of different substrates in the cycloheximide chase appears to be particularly prominent for CD147 as an ERAD substrate.The differences for various substrates should be contextualised with their overall half-lives.Given this substrate's sensitivity to SEL1L activity, experiments showing the compensatory effect of the SEL1L S658P/F668Y variant should be presented.This would substantiate the conclusions drawn from the observed partial reversal of HMW accumulation, which isn't a direct ERAD-reporting observation.We thank the reviewer for this great comment.Our data showed that SEL1L S658P/F668Y can partially rescue the defects of ERAD dysfunction towards a model substrate when overexpressed (Response Figure 4a).However, our effort to generate SEL1L S658P/F668Y knock-in HEK293T cells failed as the cells exhibited either resistance to the Cas9 electroporation or stopped growing following electroporation.In SEL1L KO HEK293T cells with overexpressed SEL1L (double) mutants, we noted that the overexpressed SEL1L, both WT and mutants, were unable to rescue ERAD dysfunction as measured by the levels of endogenous substrates in SEL1L -/-cells in vitro (Response Figure 4b).This was also true when cells transfected with nontagged SEL1L WT or mutants (Response Figure 4c).While we are currently exploring alternative approaches to address the issue raised by the reviewer, we have toned down the conclusion for this part in the revised manuscript (see the revised text below), due to the lack of data for the S658P/F668Y variant.
Results: "These data suggested that SEL1L S658P may cause a physical collision between SEL1L and HRD1 via SEL1L F668 and HRD1 Y30 residues, while having no effect on SEL1L-OS9/ERLEC1 interaction.However, the definitive support for this model will require the affinity measurements between the two proteins." Response Figure 4. Overexpressed SEL1L, either WT or mutant, is nonfunctional towards endogenous ERAD substrates in SEL1L -/- HEK293T cells.(a) Reducing and non-reducing SDS-PAGE and Western blot analyses of proAVP-G57S high molecular-weight (HMW) aggregates in WT or SEL1L -/-HEK293T cells transfected with indicated SEL1L-FLAG constructs (n = 4 independent samples for each genotype).(b-c) Western blot analysis of known endogenous substrates (IRE1α, OS9 and CD147) in WT and/or SEL1L -/- HEK293T cells transfected with indicated SEL1L-FLAG at different doses of SEL1L constructs (b) or different SEL1L constructs (c) (two independent repeats).
• Supported by the structural analysis, in Fig. 6, the authors investigate how SEL1LS658P attenuates the SEL1L-HRD1 interaction and attribute this to the interface between F668 (SEL1L) and Y30 (HRD1).While the results back their hypothesis, additional evidence would solidify this conclusion.Demonstrating that mutating HRD1, targeting Y30, produces a similar effect would be pivotal.As the co-IP method doesn't conclusively indicate the directness of interactions, these further evidences would be needed for establishing the effect on SEL1-HRD1 interaction.Affinity measurements between the variants of the two proteins would also provide a more definitive support for this core finding.We thank the reviewer for this great comment.As suggested, we have now generated HRD1 Y30A, Y30D, Y30K and Y30F mutants and assessed their impact on HRD1 interaction with SEL1L and other ERAD components.Our IP data showed that HRD1-Y30 mutated to A, D or K significantly disrupted the SEL1L-HRD1 interaction by approximately 80-90%, while Y30F retained ~ 70% interaction (Response Figure 5).The disruption of SEL1L-HRD1 interaction also abolished the interaction of HRD1 with the ER lectin OS9, while having no impact on the interaction with FAM8A1 (Response Figure 5).This data suggests that the HRD1 Y30 is critical determinant for the SEL1L-HRD1 interaction.This data is now shown in Extended Data Fig. 7c in the revised manuscript.We were not able to perform affinity measurements due to the technical challenges, while fragments of SEL1L and HRD1 are known to directly interact in vitro 4 .Hence, we now have toned down our conclusions on the direct interaction.
Result: "We first examine whether SEL1L F668 or HRD1 Y30 is critical for the SEL1L-HRD1 interaction.Indeed, HRD1 Y30 mutated to Ala (A), Asp (D) or (K) significantly disrupted the SEL1L-HRD1 interaction by 80-90%, and 30% when mutated to Phe (F).The disruption of SEL1L-HRD1 interaction abolished the interaction of HRD1 with OS9, while having no impact on the interaction with FAM8A1." Result: "These data suggested that SEL1L S658P may cause a physical collision between SEL1L and HRD1 via SEL1L F668 and HRD1 Y30 residues, while having no effect on SEL1L-OS9/ERLEC1 interaction.However, the definitive support for this model will require the affinity measurements between the two proteins." Response Figure 5. HRD1 Y30 is critical for the SEL1L-HRD1 interaction.(a) Immunoprecipitation of FLAG-agarose in HRD1 -/-HEK293T cells transfected with indicated HRD1-FLAG constructs to exam the interaction with SEL1L, OS9 and FAM8A1, with quantitation shown below the gel as means from two independent repeats.
• In Extended Data Fig. 6c, the authors suggest that the asparagine mutant (F668N) doesn't disrupt the interaction between SEL1L and HRD1.Given that asparagine isn't an aromatic residue, the authors should discuss these results in the context of their aromatic-aromatic interaction theory.