Review Papers Claim by Claim

What the paper shows (data-driven, provable claims)

• Screen: 129 purified anti‑RBD mAbs + 10 hybridoma supernatants + controls; assay used fluorogenic ACE2 substrate MCA‑APK(Dnp) in MES buffer (50 mM) with added ZnCl2, fluorescence read at Ex/Em 320/405 nm; slope (RFU/min) from early kinetics used to call activity Song et al. Methods
• Positives: 4 samples (AGRD013‑4G9, AGRD016‑2B3, AGRD016‑4H2, AGRD021‑9A10) met threshold; two of these block ACE2‑RBD binding, two enhance it; positive samples were mostly hybridoma supernatants (authors note this distinction) Song et al. Results
• Mechanistic probes: activity resisted 5 mM EDTA and was NOT inhibited by MLN‑4760 (a well‑characterized ACE2 active-site inhibitor), but for two blocking mAbs the activity was inhibited by an overlapping RBD peptide pool; kinetics give lower Km but ~10^4 fold lower catalytic efficiency (kcat/Km) than rhACE2—consistent with prior abzyme literature that abzymes are orders-of-magnitude less efficient than native enzymes Song et al. Mechanistic probes

Context — why this matters mechanistically

ACE2 is the canonical receptor for SARS‑CoV‑2 spike and is a Zn2+ metallopeptidase that cleaves Ang II to angiotensin(1‑7); structural work defines the ACE2 peptidase domain and active site, and small molecule inhibitors like MLN‑4760 bind that active site and block catalysis — so an antibody that cleaves an ACE2 fluorogenic substrate but is insensitive to MLN‑4760 and EDTA is likely using a different catalytic geometry/mechanism than native ACE2 ACE2 structure (Wrapp et al. / Yan et al.)

Critical appraisal — strengths

• Large screen of many independent mAb collections from four groups reduces chance the phenomenon is restricted to a single lab artifact Song et al. Materials
• Orthogonal controls: EDTA to remove serum ACE2 background; non‑ACE2 control substrate; peptide competition; MLN‑4760 — these interrogations begin to distinguish canonical ACE2 from abzyme-like activity Song et al. Controls

Critical appraisal — limitations, alternate explanations, and blind spots

1. Hybridoma supernatant vs purified IgG: the four positive signals came from hybridoma supernatants (authors note some positives were hybridoma-derived). Supernatant contains enzymes/proteases from cells and serum (FBS) traces; although EDTA was used to remove ACE2 background, other proteases (serine/cysteine proteases) could survive and produce substrate cleavage. The manuscript reports applying EDTA and control substrates, but identifying and fully excluding co‑purifying proteases requires additional orthogonal steps (e.g., IgG purification from the same positive supernatants followed by activity re‑measurement; size‑exclusion to confirm activity co‑elutes with IgG; proteomics mass spec of active fractions) — these were not fully performed Song et al. Limitations re: supernatants
2. Assay specificity: the fluorogenic substrate MCA‑APK(Dnp) is a commercial ACE2 probe, but small proteases with overlapping specificity can sometimes cleave peptide substrates; (i) negative control substrate was used and gave negative slopes, which supports specificity, but (ii) the best demonstration that the antibody itself (and not contaminant) is catalytic is: activity retained after extensive affinity purification of the mAb and after denaturation/renaturation tests, or detection of cleavage products on MS after incubation with pure antibody. Those rigorous confirmatory experiments are not present, so residual uncertainty remains Song et al. Substrate specificity tests
3. Inhibitor profile interpretation: the activity being EDTA- and MLN‑4760‑resistant is important — it argues the mechanism differs from zinc metalloprotease ACE2 — but it does not prove the mAbs are themselves catalytic; it could point to other proteases in the sample that are EDTA‑resistant (e.g., serine proteases) or to non‑canonical metal‑independent catalytic sites. Direct proof that the antibody variable region contains a catalytic constellation (e.g., demonstration that F(ab) fragments retain activity, or that catalytic activity maps to CDRs by mutagenesis/sequence analysis) is absent. Structural/sequence mapping (CDR sequences, Ig subclass/glycosylation, and IgFold/AlphaFold modelling) would help but is not reported Song et al. Inhibitor results
4. Low catalytic efficiency and physiological plausibility: reported kcat/Km is ~10^4 lower than rhACE2. Even if real, pathophysiologic significance requires (i) demonstration that these antibodies are produced in patients at sufficient titers, (ii) that they are stable and active in plasma/tissue microenvironments, and (iii) that their substrates (e.g., Ang II) are accessible in vivo. The paper cites prior work associating immunoglobulin‑linked ACE2 activity with blood pressure changes post‑exercise in convalescents, but direct causation in patients is still unproven Song et al. physiological significance caveat
5. Selection bias in mAb collections: many supplied mAbs were selected for neutralization and ACE2 blocking; the authors note that selection pressures in mAb isolation pipelines might bias toward rare structural motifs; therefore frequency estimates of abzyme generation in natural infection cannot be inferred from this collection. Crucially, mAbs here are not traced back to donors with Long COVID or specific symptoms Song et al. selection bias note

What experiments would move this from 'intriguing' to 'demonstrated abzyme'?

1. Affinity-purify the positive mAbs from hybridoma supernatants (Protein A/G) and re‑measure activity using the purified IgG. Show activity co‑elutes with IgG on SEC and is absent in flowthrough.
2. Generate F(ab) and Fc fragments; test whether activity resides in F(ab) (expected if catalytic) and whether deglycosylation alters activity.
3. Mass spectrometry of reaction mixtures to detect precise cleavage products and map the scissile bond — confirming identical cleavage chemistry as ACE2 vs a different site.
4. Site‑directed mutagenesis of the variable regions (CDR substitutions) or recombinant expression of the mAb V regions to demonstrate loss/gain of catalytic activity mapping to sequence motifs; use IgFold/Alphafold/ESM modelling to design mutants.
5. Proteomics of active vs inactive hybridoma supernatants to identify co‑purifying proteases; include protease family inhibitors (serine, cysteine inhibitors) to profile sensitivity.
6. Measure activity in patient plasma with known titers of the clone (if traceable), and perform in vivo or ex vivo assays to test impact on Ang II levels and downstream physiology.

These steps would exclude contaminants, localize catalysis to antibody sequences, and show physiological relevance.

Judgment summary (evidence-weighted)

• Evidence the samples cleave an ACE2 fluorogenic substrate in vitro: present and replicable in the authors' hands (kinetics, peptide competition) Song et al. kinetic evidence
• Evidence that the antibody molecule (variable region) is the catalytic species: insufficient — purification, proteomics, F(ab) tests, cleavage product MS, or sequence-to-activity mapping are lacking.
• Mechanistic conclusion that these are ACE2‑like abzymes with in vivo relevance: hypothesis (plausible but unproven). The data are consistent with catalytic antibodies (abzymes) but do not yet exclude alternative causes.

Actionable next steps — experimental pipeline (concise)

1. Affinity‑purify positives → SEC fractionation → activity assay on each fraction → LC‑MS/MS proteomics of active fraction.
2. Produce recombinant IgG (sequence the hybridoma V regions if not available), express as recombinant human IgG1, test activity vs parent hybridoma supernatant.
3. Map cleavage site(s) on physiological substrates (Ang II) by MS to compare with ACE2 cleavage products.
4. Mutagenize CDRs predicted by IgFold/ESMFold to identify catalytic residues; test recombinant mutants.
5. If antibody is catalytic, test plasma-level prevalence in cohorts (acute, convalescent, long COVID) and correlate with physiological readouts (BP, Ang II/Ang(1‑7) ratio, coagulopathy markers).

Key references (supporting context)

• Song et al. 2026 (primary)
• ACE2 structure (2020)
• Tramontano et al. 1986 (abzymes)