BGPT: Paper Review: Dynamic O-GlcNAcylation of Sec23-interacting protein regulates COPII function

Fuel Your Discoveries

Quick Explanation Copied

Core claim (mechanism)

The paper argues that dynamic, site-specific O-GlcNAcylation of Sec23-interacting protein (Sec23IP) tunes COPII by promoting Sec31A recruitment early (via constitutive IDR glycosylation) and reducing Sec31A binding later (via increased O-GlcNAc during cargo transport), thereby fine-tuning ERES assembly/uncoating.

Long Explanation

Paper review (skeptical, evidence-based)

Title: Dynamic O-GlcNAcylation of Sec23-interacting protein regulates COPII function

What the authors set out to test

Is Sec23IP actually O-GlcNAcylated in cells?
Which Sec23IP function depends on O-GlcNAc? (test with a nearly unglycosylatable Sec23IP IDR mutant).
Does O-GlcNAc tune Sec23IP interactions with COPII components (Sec31A vs Sec23A)?
Is Sec23IP O-GlcNAc dynamic during cargo transport?

Mechanism map (authors’ model)

Blue = claims supported by the paper’s experiments; orange = model language that requires residue-level confirmation.

Confidence tags

Moderate: Sec23IP is O-GlcNAcylated; NA reduces Sec31A binding; O-GlcNAc increases over transport while Sec31A interaction decreases.
Speculative: residue-specific assignment of the exact glycosites that drive early vs late phases, and the claim that later O-GlcNAc promotes uncoating specifically. The paper itself flags residue identification as a future task.

Evidence appraisal (step-by-step)

1) Sec23IP O-GlcNAcylation is experimentally validated

O-GlcNAcylated Sec23IP is detected by multiple monoclonal antibodies (18B10, RL2, 9D1), and signals shift in expected directions with OGT inhibition (5SG) vs OGA inhibition (TG).
Antibody specificity is supported by CRISPR KO controls (no Sec23IP signal in KO IPs).

Skeptical note: The paper validates using antibody recognition, not residue-level MS mapping in purified endogenous Sec23IP in this work. The authors acknowledge that full site identification will require future systematic MS on purified endogenous protein.

2) O-GlcNAc on the Sec23IP IDR is functionally required for conventional ER exit

E-cadherin transport (RUSH) is delayed in Sec23IP KO HeLa cells compared to WT.
A nearly unglycosylatable Sec23IP IDR mutant (NA; mutations across nearly all predicted glycosites in the N-terminal IDR) localizes normally to ERES (via Sec16A colocalization) but produces increased Sec23IP and Sec16A puncta (consistent with ER exit-site accumulation).
Rescue in Sec23IP KO cells: WT Sec23IP restores E-cadherin RUSH, whereas NA gives only partial rescue with reduced Golgi-localized signal.

Skeptical note: The NA mutant may not eliminate all physiological O-GlcNAc sites (residual immunoreactivity is noted), so phenotypes could reflect both incomplete loss and broader changes in IDR sequence/charge context from mutagenesis.

3) Glycosylation tunes Sec23IP–Sec31A, not Sec23IP–Sec23A

Co-IP shows Sec23A binding is similar for WT vs NA, while Sec31A binding is significantly reduced for NA.
IF corroborates recruitment: Sec23IP KO reduces Sec31A puncta MFI despite normal Sec31A expression; WT rescue restores Sec31A puncta while NA does not.

Mechanistic blind spot: The paper’s discussion suggests IDR O-GlcNAc sites lie outside the previously reported Sec31A-interacting region, implying an indirect mechanism (altered Sec23IP structure/affinity or multivalent/condensate effects). That indirect biophysical basis is not directly tested here (e.g., no LLPS assays, no structural readout).

4) Sec23IP O-GlcNAc increases during cargo transport and correlates with loss of Sec31A interaction

During RUSH E-cadherin trafficking (timepoints including ~20 min), Sec23IP O-GlcNAc measured with 9D1 increases; RL2 shows a similar trend but not statistically significant; 18B10 stays largely unchanged.
Co-IP reveals Sec23IP–Sec31A interaction decreases markedly at ~20 min, coinciding with increased glycosylation; Sec23IP–Sec23A interaction remains unaffected.
IF indicates Sec31A distribution is stable, while Sec23IP puncta intensity increases and puncta number decreases during transport.

Skeptical note: Correlation is clear, but causality for the “uncoating” interpretation is not directly demonstrated (no direct measurement of uncoating kinetics, Sec23/Sec31 dissociation timing, or cargo-loaded vesicle maturation state).

Context within known biology (only what’s supported)

COPII is built from inner Sec23/24 and outer Sec13/31 layers and initiates assembly at ER exit sites. and .

O-GlcNAc is a reversible intracellular modification installed by OGT and removed by OGA, often used to relay nutrient/state information into protein behavior. and .

Limitations / potential confounds (what could mislead)

Residue-level uncertainty: NA mutates predicted IDR glycosites, but residual O-GlcNAc signal suggests additional sites; exact inducible vs constitutive residues are not mapped.
Antibody epitope bias: Different antibodies (9D1/RL2/18B10) show different timecourse responses, implying epitope/site specificity; interpretation depends on how these epitopes map onto individual glycosites.
Overexpression / rescue system artifacts: Rescue constructs and tagged proteins can perturb localization or stoichiometry; phenotype strength may differ with expression level and in other cellular contexts. The paper itself relies mainly on cell culture models.
Opposing cargo conclusions: Sec23IP KO affects E-cadherin (conventional) transport but not collagen transport in SW1353 under the tested assay conditions, raising the possibility of cargo- and context-dependent requirements or compensation.
“Uncoating” not directly measured: The model links inducible glycosylation to uncoating, but the experiments primarily quantify glycosylation, co-IP interactions, and puncta; direct uncoating kinetics or structural changes at single-ERES level are not provided.

How to falsify the paper’s core mechanism (most discriminating tests)

Site-rescue: If you mutate only the putative constitutive glycosites (and leave inducible ones) and can fully rescue Sec31A recruitment/early transport, that supports the “early glycosites recruit Sec31A” idea; failure would weaken it.
Temporal causality: If pharmacologically altering O-GlcNAc cycling changes Sec31A binding dynamics in the same direction and timing predicted by the RUSH timecourse, that would support causality; if it changes global glycosylation but not Sec31A dynamics, the link is weaker.
Direct uncoating measure: If later-stage glycosylation does not change measurable uncoating readouts (e.g., Sec31A dissociation kinetics from ERES in real time), the uncoating part of the model falls apart even if Sec31A binding correlations hold.

Useful BGPT follow-ups (author-focused & mechanism-focused)

Author reviews (click to explore each author’s related work)

Feedback:

Updated: April 14, 2026