Review Papers Against Their Own Results

Visual — Key numeric results (paper-derived)

Visual — Age & vector-design dependence

Explain — Short, evidence-first critique

1. What they did (concise): Systemic IV AAV8 delivery of SpCas9-compatible gRNAs (ssAAV and scAAV formats; single or multi-copy gRNAs) into Ai9;SpCas9-EGFP and Townes;SpCas9-EGFP mice to read out in situ editing of HSCs by tdTomato activation and by amplicon sequencing of disease locus (HBG1), with functional readouts (CFU, NSG transplantation, HbF by flow/HPLC) and toxicity checks (Main methods and approach
2. Key quantitative findings:
   • Adult editing: ssAAV‑1x at 5×10^12 vg → ~0.6% tdTomato+ HSCs; scAAV‑1x → ~3.7%; scAAV‑2x → ~5.8% (means ± SD; multiple cohorts) (Adult editing efficiencies)
   • Neonatal editing: optimal ssAAV‑1x at 1×10^12 vg → ~1.4% HSCs; scAAV‑2x at same neonatal regimen → ~22% tdTomato+ HSCs (Neonatal editing efficiencies)
   • Townes disease locus: scAAV‑3xHBG1.gRNA produced high % modified reads in sorted adult HSCs (A/A genotype avg ~56%; S/S ~40%) and significant editing in BM erythroid progenitors and blood; HbF increased in S/S mice but no phenotypic rescue observed (Townes HBG1 editing and HbF)
3. Strengths — why this matters:
   • Careful comparative vector-design experiments identify practical parameters (scAAV, higher gRNA:Cas9 ratio) to raise in situ editing — directly actionable for other labs (Vector design comparisons).
   • Functional validation: edited HSCs retain CFU potential and long-term engraftment in NSG recipients — critical for safety/utility claims (Functional assays (CFU, transplant)).
4. Key limitations & open questions (critical):
   • Off-targets and large on-target deletions not comprehensively profiled. The paper uses amplicon NGS (CRISPResso2) to quantify editing at intended loci but does not present long-read sequencing, genome-wide off-target discovery (GUIDE-seq, CIRCLE-seq), or single-cell outcome maps that reveal large deletions, translocations, or vector-genome integrations; these classes are now known to be frequent in vivo and cell-type dependent (In vivo repair outcomes and large on-target events).
   • AAV-associated toxicity at high doses and DDR interactions. The paper notes tolerability up to certain vg/kg thresholds but also reports dose-dependent toxicity in neonates at very high doses; independent work shows that rAAV ITRs can bind and sequester DDR proteins (e.g., PARP1) causing replication stress and toxicity in dividing cells — especially relevant for neonatal/proliferative HSCs and for high systemic doses (AAV ITR–DDR interactions and toxicity).
   • Translational gaps: mouse HSC biology, AAV8 tropism in mice, and Cas9-transgenic models differ from human HSCs and clinically relevant delivery constraints (preexisting anti-AAV antibodies, scale of vg/kg, human bone-marrow microenvironment). The field increasingly documents strong cell-type and species-specific differences in repair outcomes and tropism, so mouse numbers may overestimate safety or efficacy in humans (AAV tropism atlas (mouse)).
   • Dose- vs biology-driven effects confounded: Neonates showed higher editing partly due to higher vg/kg but also to developmental biology differences; authors model dose below 1.5×10^15 vg/kg but caution about toxicity at extreme doses — this complicates prediction for humans where safe systemic vg/kg limits are lower.
5. Practical recommendations for follow-up experiments:
   1. Perform long-read (ONT/PacBio) capture sequencing of on-target sites from sorted HSCs to quantify large deletions, insertions, and vector-genome junctions (AAV-LI/Endo-LI) as in recent in vivo repair studies (Need for long-read on-target SV detection).
   2. Run genome-wide off-target discovery (GUIDE-seq / CIRCLE-seq / DISCOVER-seq) in primary HSPC-like cells (human CD34+) with the exact gRNA sequences to quantify off-target risk and enable high-fidelity guide selection.
   3. Assess AAV ITR effects on proliferating HSCs (PARP1 binding, replication stress markers) at the doses used and test ITR-engineered/attenuated capsids if toxicity is observed (ITR–DDR toxicity and mitigation).
   4. Translate to human HSPCs using ex vivo human CD34+ xenograft/NSG or NBSGW models with matched systemic AAV exposure to test delivery and editing outcomes with human cells in vivo, and quantify functional rescue (e.g., HbF, or HBB correction) and long-term clonal dynamics (Human HSPC HDR xenografts as translational benchmark).
6. Confidence and how evidence could overturn conclusions:
   • Current claims that AAV-gRNA can edit HSCs in situ and that scAAV and increased gRNA copy number improve efficiency are well-supported by the dataset and reproducible within the mouse system (Support for main claims).
   • However, the clinical-translation conclusion (that AAV-CRISPR in vivo is readily translatable to humans) remains provisional: demonstration of widespread large deletions, vector integrations, DDR toxicity, or adverse clonal dynamics in human-relevant models would materially change the risk/benefit calculus and could falsify optimistic translational interpretations (see references above on in vivo repair complexity and AAV ITR effects) (Potential falsifiers: large events).

Methods transparency & reproducibility

Methods are reported in detail (vector production cores, exact serotype AAV8, injection routes, immunophenotyping gating strategies, CRISPResso2 for amplicon analysis, CFU/transplant protocols), which supports reproducibility across labs; however:

• No public deposition of raw sequencing data or long-read capture libraries is provided in the manuscript (data availability not stated), which reduces external reanalysis potential. Raw amplicon/NGS reads and vector copy-number tables should be deposited to SRA/GEO to maximize reproducibility (Methods completeness and data availability).

Final appraisal (summary)

This is a high-quality, carefully controlled preclinical study that meaningfully advances the concept that AAV-delivered CRISPR gRNAs can edit HSCs in situ and identifies practical vector design rules (scAAV; higher guide copy) that increase efficiency. The demonstration of preserved HSC function after in situ editing is an important positive safety signal. Nevertheless, before clinical translation, the field needs long-read and genome-wide analyses to quantify large on-target rearrangements and vector integrations, explicit characterization of AAV-ITR–DDR interactions at therapeutic doses, and validation in human HSPC xenograft contexts. The study is a valuable step but not yet a translational finish line.