SIRT2 and "defense response to virus" (GO:0051607): Restriction vs. Dependency

Gene: SIRT2 (human) · UniProt: Q8IXJ6 · Taxon: NCBITaxon:9606
Focus: function_assignment · Seed hypothesis: SIRT2 has defense response to virus (GO:0051607)
Deliverable: direction verdict + decisive experiments + contrary evidence

Executive Judgment

Verdict: REFUTED as an unqualified assignment / OVER-ANNOTATION. Direction is VIRUS-SPECIFIC, and the preponderance of decisive perturbation evidence points to SIRT2 acting as a host DEPENDENCY (pro-viral) factor — the opposite sign to GO:0051607.

A single, unqualified "defense response to virus" annotation on human SIRT2 is not warranted. The sign of SIRT2's effect on virus replication is not uniform, and where the readout is infectious progeny titre (the decisive metric), SIRT2 behaves as a factor that viruses require, not one that restricts them:

• Both decisive infectious-progeny studies point pro-viral. In dengue, a SIRT2-directed inhibitor reduces release of infectious progeny and the effect is abolished by SIRT2 knockdown (on-target; PMID 34147476). In HIV-1, a SIRT2 inhibitor reduces viral growth by >1 log₁₀ in macrophages and humanized mice (PMID 41883165).
• HBV (6 studies, multiple perturbation types) consistently shows SIRT2 promotes replication: overexpression increases HBV nucleic acids (PMID 35663860), and three independent chemical inhibitors/modulators (AGK2 — PMID 30275764, 40270769; FLS-359 — PMID 38641024) suppress HBV.
• HSV-1: SIRT2 suppresses the antiviral cGAS-STING/type-I-IFN axis, so SIRT2 loss or inhibition increases interferon and clears virus (PMID 37870259) — again pro-viral.

Contrary evidence exists but is weaker for GO:0051607. The antiviral (restriction-like) direction rests on (i) influenza, where SIRT2 activation lowers replication via redox restoration (PMID 35071051) — an activation gain-of-function with a surrogate readout, not loss-of-function; (ii) SIRT2 support of IFN-stimulated-gene transcription via CDK9/STAT1 (PMID 30487288) — a signaling surrogate, no viral titre; and (iii) a minor alternatively spliced isoform 5 that inhibits HBV (PMID 32493816) — isoform-specific and opposite to the dominant isoform 1.

Key caveat / important curation fact: A QuickGO query of all 206 GO annotations on Q8IXJ6 found no existing viral-defense annotation (GO:0051607 and all related/descendant viral terms return 0 hits). The seed is therefore a proposed addition, and the evidence does not support adding it in unqualified form.

Evidence Matrix

Tally: 10 refute (pro-viral/dependency), 3 support (all activation-based, signaling-surrogate, or minor-isoform), 1 qualifies (anti-inflammatory protection, not restriction). Provenance: evidence-matrix code + output executed in Iteration 2; CSV saved to the code sandbox at /tmp/sirt2_virus_evidence_matrix.csv.

GO Curation Implications (leads requiring curator verification)

• GO:0051607 (defense response to virus, BP): Do NOT add as an unqualified annotation. It is not currently in the GO record for Q8IXJ6 (QuickGO: 0 hits), and the primary literature does not support a uniform host-restriction role. If any viral statement is captured at all, it should be virus-qualified and directionally correct.
• The dominant, mechanistically coherent signal (HBV, dengue, HIV-1, HSV-1) is that SIRT2 is pro-viral / permissive. If a curator wishes to record this, the accurate directional term is closer to positive regulation of viral genome replication (GO:0045070) or negative regulation of type I interferon-mediated / innate immune response (e.g., via GO:0032480 negative regulation of type I IFN production), each paper- and virus-specific with the appropriate NOT/qualifier discipline — not GO:0051607.
• Preferred core annotations remain the well-supported ones: NAD⁺-dependent protein/α-tubulin deacetylase activity (MF), protein deacetylation (GO:0006476, BP, IDA), and cytoplasm/microtubule/centrosome localization (CC). The antiviral phenotypes are downstream, context-dependent consequences of this deacetylase activity, not a distinct "defense response to virus" function.
• Isoform caution: Any HBV-restriction statement (PMID 32493816) is isoform-5-specific and should not be propagated to the canonical Q8IXJ6 entry.

Mechanistic Scope

The immediate molecular function being modulated in every study is SIRT2's NAD⁺-dependent lysine deacetylase activity (substrates include α-tubulin, G3BP1, CDK9, G6PD, p53, β-catenin pathway components). Virus outcomes are downstream of this enzyme activity:

• Pro-viral routes: SIRT2 deacetylates G3BP1 to disassemble the cGAS-G3BP1 condensate and dampen STING/IFN (HSV-1, COVID); supports AKT/GSK-3β/β-catenin and p53-dependent enhancer activity for HBV; is co-opted by dengue and HIV-1.
• Antiviral routes: SIRT2 deacetylates CDK9 to promote STAT1-Ser727/ISG transcription; deacetylates G6PD to sustain NADPH/GSH redox that limits influenza.

Thus "defense response to virus" is at best a pleiotropic, context-specific downstream phenotype, not a primary gene-product function.

Conflicts and Alternatives

• Genuine biological heterogeneity, not artifact: same enzyme, opposite outcomes depending on which acetyl-substrate/pathway dominates in a given virus/cell system.
• Perturbation-type asymmetry: antiviral effects rely on SIRT2 activation or signaling surrogates; pro-viral effects are shown by inhibition + genetic knockdown with on-target rescue (dengue) and in vivo inhibition (HIV-1) — methodologically stronger for settling the sign.
• Isoform confounding: isoform 1 (pro-HBV) vs isoform 5 (anti-HBV) from the same locus.
• Paralog caution: many "sirtuin" antiviral claims involve SIRT1 or pan-sirtuin tools (e.g., tenovin-1 is SIRT1/2; resveratrol/SRT compounds activate SIRT1). Curators must avoid SIRT1→SIRT2 carry-over.
• Phenotype vs restriction: SARS-CoV-2 protection (PMID 40220296) is reduced immunopathology/disease severity, not reduced viral replication — a different GO concept from GO:0051607.

Knowledge Gaps

1. Clean genetic loss-of-function on infectious titre. Most decisive HBV/HIV data use inhibitors; only dengue pairs inhibitor with siRNA. Checked: no CRISPR-KO study reporting infectious titre increase upon SIRT2 loss. Resolve: isogenic SIRT2-KO vs WT, plaque/TCID50 across virus panel.
2. Catalytic dependence. Checked: inhibitor data imply catalytic requirement but few catalytic-dead (H187Y) rescues. Resolve: WT vs catalytic-dead re-expression in KO.
3. Isoform-resolved effects. Resolve: isoform-specific knockdown/expression with titre readout.
4. Inhibitor off-target / SIRT1 contribution. Resolve: SIRT2-selective genetic tools alongside chemistry.
5. Whether any virus is truly restricted by endogenous SIRT2 (LoF ↑ titre). Checked: not found for influenza (only activation GoF). Resolve: SIRT2 depletion + IAV plaque assay.

Discriminating Tests

• Isogenic SIRT2-KO (CRISPR) vs WT, infectious-progeny titre (plaque/TCID50), across DENV, HIV-1, HBV, HSV-1, IAV — the single most decisive experiment for the sign.
• Catalytic-dead rescue (H187Y) to tie any effect to deacetylase activity.
• SIRT2-selective inhibitor vs pan-sirtuin tool side-by-side to exclude SIRT1.
• Isoform-specific reconstitution (isoform 1 vs 5) in KO cells.
• G3BP1/cGAS-STING reporter ± SIRT2 to confirm the immunosuppressive (pro-viral) axis quantitatively.

Curation Leads (require curator verification)

• Action: Reject/withhold unqualified GO:0051607 on Q8IXJ6 (over-annotation). If a viral role is captured, make it virus-specific and directionally correct (predominantly pro-viral / negative regulation of innate antiviral response).
• Candidate references + snippets to verify:
• PMID:34147476 — "the release of progeny viruses was inhibited by tenovin-1 treatment" and "The antiviral effect of tenovin-1 was abrogated with SIRT agonist or SIRT2 knockdown treatment, which indicated the effect of tenovin-1 was on-target" (decisive, pro-viral).
• PMID:41883165 — "Sirtinol, a Sirt2 inhibitor, significantly reduced HIV-1 growth in M0, M1, and M2-MΦs by >1log10 over 7 days" (pro-viral, in vivo).
• PMID:35663860 — "ectopic SIRT2 overexpression markedly increased total HBV RNAs, 3.5-kb RNA and HBV core DNA" (pro-viral).
• PMID:37870259 — SIRT2 suppresses cGAS-STING; "SIRT2 deficiency markedly elevated the expression levels of type I interferon" (pro-viral/immunosuppressive).
• Contrary (to weigh, not to base GO:0051607 on): PMID:35071051 (IAV, activation), PMID:30487288 (ISG signaling), PMID:32493816 (isoform 5).
• Candidate replacement/qualified terms: positive regulation of viral genome replication (GO:0045070); negative regulation of type I interferon production (GO:0032480) / negative regulation of innate immune response — each paper- and virus-scoped. Retain core MF (protein/tubulin deacetylase) and CC (cytoplasm/microtubule).
• Suggested curator questions: Is the proposed GO:0051607 from text-mining/paralog inference? Does any cited paper show LoF increasing infectious titre? Are isoform and SIRT1 confounds controlled?

Limitations

• Assessment is literature- and database-driven (PubMed + QuickGO); no wet-lab replication.
• Several HBV/HIV readouts are surrogates (nucleic acid, growth) rather than plaque titre.
• Literature coverage, while broad (HBV, HIV-1, dengue, HSV-1, IAV, SARS-CoV-2), is not exhaustive; additional virus-specific reports may exist.

Direction verdict: Virus-specific, dependency-dominant. Decisive infectious-progeny evidence (dengue PMID 34147476; HIV-1 PMID 41883165) and the HBV/HSV-1 bodies of work place SIRT2 as a host dependency / pro-viral factor, contradicting an unqualified "defense response to virus" assignment. Restriction-like observations are real but confined to specific contexts (influenza via activation; IFN/ISG signaling; a minor HBV isoform) and do not justify GO:0051607 for canonical human SIRT2.

Artifacts

• OpenScientist final report
• OpenScientist final report