BGPT: Paper Review: Cryo-EM structure of human PAPP-A2 and mechanism of substrate recognition

Explore by Goal

Quick Explanation Copied

What this cryo-EM paper adds

PAPP-A2 is structurally resolved (apo; 3.13 Å) for the N-terminal LG, MP, and M1 domains, and it behaves as a monomer in the reconstructed density ensemble.
The authors propose that PAPP-A2 shares a similar IGFBP5 anchor-peptide recognition mode with PAPP-A, but with substantially reduced IGFBP5 cleavage efficiency.
The mechanistic centerpiece is a dynamic / “open” M2 domain (supported by MD simulations + functional/biochemical tests) that likely reduces productive anchor engagement and explains both lower cleavage and the disease mutation A799V’s loss-of-activity.

Source paper:

Long Explanation

Cryo-EM structure of human PAPP-A2 & substrate recognition — skeptical paper review

Paper:

Research question (as framed by authors)

How does human PAPP-A2 recognize and cleave IGFBP5 (and relate to paralog PAPP-A), and how does the disease-associated mutation A799V abolish PAPP-A2 proteolytic activity?

Evidence base used: cryo-EM (apo), functional proteolysis assays, binding assays, thermal stability (NanoDSF), and ML/MD mechanistic inference.

Key quantitative deltas between PAPP-A2 vs PAPP-A (from the paper)

Fold-changes reported in the manuscript (inactive mutants for binding; active enzymes for cleavage). These are direct, paper-stated comparative metrics.

Interpretation: The paper’s central claim—reduced IGFBP5 cleavage by PAPP-A2—has two measurable components: weaker binding (in the binding assay setup using catalytically inactive constructs) and lower catalytic turnover/cleavage efficiency when active enzymes are compared under equimolar enzyme conditions.

Source:

Structural resolution limits and what’s actually known

The cryo-EM reconstruction is high-resolution for the regions it resolves (3.13 Å), but the authors explicitly report missing density for several domains (including the M2 domain, CCP1-5, and LNR3). This matters because the paper’s mechanistic model for IGFBP5 recognition is partly inferred from ML+MD rather than directly observed in a substrate-bound cryo-EM complex.

Resolved: N-terminal LG, MP, M1 domains (model built for residues 25–926; ambiguous density for LNR1/2 segment 373–420).
Unresolved / flexible: M2, CCP1–5, LNR3 (authors suggest flexibility and/or lack of substrate complex occupancy during sample prep).

Source:

MD-derived geometry supporting “open” M1–M2 separation (reported summary statistics)

The authors quantify domain separation distances in MD as a mechanistic proxy for M2 “openness” and for the anchor peptide’s ability to engage M2. Below are the reported summary statistics for their main CHARMM36m simulations.

Skeptical interpretation: Because cryo-EM does not directly visualize the M2 domain and does not directly resolve the substrate-bound complex, these MD distances are plausibly informative but still model-dependent (starting structure from ML prediction; force field choice and sampling assumptions can shift domain ensembles).

Source:

A799V: functional loss + thermal destabilization (reported directionality)

The patient mutation A799V abolishes IGFBP3/IGFBP5 cleavage and shows a modest thermal stability reduction (~2 °C lower Tm) by NanoDSF.

Importantly, the mutation is >26 Å from the catalytic zinc in their mapped structure; the authors propose long-range interaction networks that stabilize activity-critical patches.

Source:

Mechanistic claims: what is strong vs what is inferred

Strongly supported (within the paper’s own experimental scope)

PAPP-A2 exists as a monomer in their cryo-EM reconstruction and in biochemical assays (reducing/non-reducing SDS-PAGE, SEC behavior, native PAGE), contrasting with dimeric PAPP-A.
Catalytic-site and activity patches matter: mutational disruption of the metzincin motif region and specific residues in the anchor-peptide-related groove abolishes cleavage for IGFBP5/IGFBP3 as measured by the authors’ cleavage assays.
A799V loss-of-function is reproducibly associated with reduced cleavage and modest destabilization under the authors’ purification and assay conditions.

More inferred / model-dependent (epistemic humility)

Substrate-bound geometry for PAPP-A2: cryo-EM did not yield observable IGFBP5 density in their prepared substrate-bound attempt (using an inactive E500Q background), so the binding pose is primarily inferred via ML modeling (AlphaFold-multimer) rather than directly seen in density.
“Open M2” ensemble: the MD-based open conformations are produced from a starting structure derived from ML complex prediction; domain separation and anchor peptide distances can depend on (i) initial model accuracy, (ii) force-field realism, and (iii) sampling adequacy. The authors partially address this by running alternative force fields and multiple replicates, but the mechanistic inference still does not replace a directly observed substrate-bound experimental structure.

Blind spots / limitations / “what would change my mind?”

Complex occupancy risk: their inability to see substrate density suggests dissociation during prep. A mechanism based on a substrate pose not directly observed could be correct in solution but still quantitatively off; this affects the strength of claims about the exact anchor-to-M2 interaction geometry.
Equimolar comparison assumptions: cleavage/binding comparisons hinge on correct interpretation of “active enzyme concentration” and monomeric/dimeric states in each assay context; the paper argues against dimerization for PAPP-A2, but assay-specific functional competence still depends on purification quality and construct integrity.
ML local confidence (pLDDT) uncertainty: the authors note relatively low confidence for the anchor peptide in the ML model; this supports caution that predicted contact residues might be partially mispositioned even if mutagenesis at some positions yields consistent phenotypes.
In vivo regulatory complexity: the paper discusses IGF-axis regulation and inhibitors (e.g., stanniocalcins, proMBP) and emphasizes that PAPP-A2 regulation mode in serum remains unclear; in vivo cleavage efficiency likely depends on additional factors not captured in their recombinant assays.
Falsification targets: the paper’s key mechanistic bet is “dynamic open M2 reduces productive anchor engagement.” A decisive disproof would require (i) experimental evidence showing M2 is actually stabilized in a substrate-bound productive state despite the cryo-EM missing density, or (ii) mutations that specifically reduce M2 dynamics without perturbing the anchor binding groove do not restore cleavage.

Source (for these limitations’ grounding in how the study was done):

BGPT usefulness for next steps (actionable)

If you’re using this paper to plan experiments/computation, the most directly actionable “interface” is the set of activity-critical patches and anchor-contact residues that the authors connect via mutagenesis + cleavage outcomes, and the mechanistic hypothesis that modulating M2 dynamics alters cleavage efficiency.

Author reviews (bespoke BGPT pages)

Links to author-named BGPT review pages for the paper’s full author list.

Feedback:

Updated: April 24, 2026