Anti-CRISPR Proteins

Phage-encoded inhibitors of bacterial CRISPR-Cas — an evolutionary arms race

Chris Mungall | AI-Assisted Gene Review

2026-06-22

Why anti-CRISPR proteins?

Anti-CRISPR (Acr) proteins are phage-encoded inhibitors of bacterial CRISPR-Cas immune systems — a striking example of the evolutionary arms race between bacteria and their viral predators.

They present four distinct challenges for GO annotation:

  1. Novel molecular mechanisms — often not well-captured by existing GO terms
  2. Dual-targeting strategies — some Acrs interact with both protein and RNA components
  3. Regulatory complexity — expression tightly regulated via Aca (anti-CRISPR associated) repressors
  4. Sparse annotations — many Acr proteins carry only generic IEA annotations

Mechanisms are well-studied structurally, but annotations lag behind.

Key biology: Acr and Aca

Acr proteins inhibit CRISPR-Cas, organized by the CRISPR type they target:

  • Type I-F (AcrIF) — target the Csy surveillance complex
  • Type I-E (AcrIE) — target the Cascade complex
  • Type II (AcrIIA / AcrIIC) — target Cas9 / Cas12
  • Type III — target Csm / Cmr complexes

Aca repressors regulate Acr expression in a negative-feedback loop.

Inhibition strategies

Acr proteins disable CRISPR-Cas through diverse mechanisms:

  • Blocking DNA recognition
  • Preventing R-loop formation
  • Mimicking DNA substrates
  • Direct crRNA binding (a unique mechanism)
  • Enzymatic inactivation

This mechanistic diversity is exactly what makes precise GO annotation hard.

The approach: AI-assisted gene review

We apply the AI gene review framework to Acr proteins:

  • Review existing GO annotations against strict GO guidelines
  • Synthesize literature + structural biology + bioinformatics
  • Identify over-general or missing annotations
  • Propose specific replacement and new GO terms where the ontology is insufficient

Acrs are excellent targets: structurally well-characterized, but under-annotated.

Pectobacterium phage ZF40 · Organism BPZF4 · UniProt H9C181 · Status COMPLETE

  • 92-amino acid protein inhibiting the Type I-F CRISPR-Cas system
  • Dual-targeting: binds BOTH the Cas7f protein backbone AND the crRNA scaffold
  • Contacts crRNA nucleotides U[+21], U[+22], G[+23] at <4 Å distance
  • Blocks R-loop formation and prevents target DNA recognition
  • Cryo-EM structure at 3.42 Å (PMID:32170016)

AcrF8: annotation issues identified

Existing term Problem Action
GO:0052170 — symbiont-mediated suppression of host innate immune response Too general MODIFY → GO:0098672
GO:0098672 — symbiont-mediated suppression of host CRISPR-cas system Specific target Replacement term
Ribonucleoprotein complex binding Missing Add core function

A more specific term (GO:0098672) precisely captures the CRISPR-Cas target,
and the dual protein-RNA binding requires explicit core-function annotation.

AcrF8: proposed new GO term

Name: CRISPR RNA binding anti-CRISPR activity

Justification: Current terms do not distinguish between Acrs that only bind Cas proteins and those that directly contact crRNA.

AcrF8 defines a unique class with dual protein-RNA binding:

  • It contacts crRNA nucleotides at <4 Å
  • Existing GO terms cannot express this mechanism

A dedicated term would let curators capture the crRNA-binding mechanism distinctly.

Key mechanisms to annotate

  1. Surveillance complex binding — GO:0043021 (ribonucleoprotein complex binding)
  2. CRISPR-Cas suppression — GO:0098672 (symbiont-mediated suppression of host CRISPR-cas system)
  3. DNA mimic function — where Acrs structurally mimic DNA
  4. crRNA binding — direct RNA contacts (needs a new term?)

Challenges of GO annotation for Acrs

  • Mechanistic novelty outpaces the ontology — terms for crRNA-binding inhibition do not yet exist
  • Dual targeting (protein + RNA) is hard to express with a single MF term
  • Generic IEA annotations dominate, leaving rich structural knowledge uncaptured
  • Regulatory context (Aca feedback loops) adds layers not reflected in current annotations
  • Over-general terms (e.g. GO:0052170) need MODIFY actions to more specific targets

Key references

  • Bondy-Denomy J et al. (2013) Nature — Anti-CRISPR discovery
  • Pawluk A et al. (2016) Cell — AcrIF mechanisms
  • Wang J et al. (2020) Nat Commun — AcrF8/F9/F6 cryo-EM structures (PMID:32170016)
  • Stanley SY et al. (2019) Cell — Aca repressor mechanisms (PMID:31474367)

Presented at the Gene Ontology Consortium Meeting, October 2025, Cambridge UK.

Conclusions & future directions

Status: project COMPLETE for the AcrF8 flagship review

The AcrF8 example demonstrates:

  1. Need for more specific GO terms for CRISPR-Cas inhibition mechanisms
  2. Value of structural biology in informing function annotations
  3. Importance of distinguishing protein-only vs. protein+RNA binding mechanisms

Next steps:

  • Identify additional Acr proteins in UniProt / QuickGO
  • Review Aca regulator annotations
  • Propose new GO terms for crRNA-binding mechanisms