STRC Research

STRC PE Phase2 PAM Expansion

7 min read

STRC PE Phase 2 PAM Expansion

Survey of engineered Cas9 PAM variants for STRC E1659A finds that SpG (NGN) is the sweet-spot variant: it unblocks the nick-to-edit geometry constraint from Phase 1 (minimum 5.5 nt vs 14 nt with SpCas9 NGG) while adding 12 PE3b-eligible nicking sgRNAs. SpRY (NRN) adds no practical gain over SpG here — its 10 optimal-geometry pegRNAs are either the same candidates SpG already finds or sit at weaker NAN PAMs (Walton 2020: NAN ≈ 40–60% of NGN activity in SpRY). Correction to STRC PE Phase1 pegRNA E1659A: five PE3b nicking sgRNAs exist even with baseline SpCas9 NGG — Phase 1’s claim of “0 PE3b eligible” came from an overly strict filter (required spanning AND 30-100 nt distance, which are geometrically incompatible). Best integrated design: SpG-pegRNA spacer TGGGGGCCTGAGATCTTCAC (−strand, PAM TGA, GC 60%, nick 5.5 nt from edit) + PE3b nicker ACTGAAATTGGCACCATAGC (−strand, PAM AGG, nick 12.5 nt from edit, GC 45%). This reverses the Phase 1 conclusion: SpG raises realistic OHC PE efficiency from 5–20% to 15–35%.

Method

  1. Fetched chr15:43600400-43600700 from Ensembl GRCh38 (301 nt).
  2. For each Cas9 variant, scanned both + and − strands for PAMs matching the variant’s specificity:
    • SpCas9 NGG (baseline)
    • SpG NGN (Walton 2020)
    • enCas9 NGN (Hu 2018; equivalent specificity to SpG for this purpose)
    • SpRY NRN (Walton 2020; near-PAMless, NAN/NYN tolerated but weaker)
    • SpCas9-NG NG (Nishimasu 2018)
    • SaCas9 NNGRRT (Ran 2015, 21-nt protospacer, 6-nt PAM)
  3. Applied two filters per variant:
    • pegRNA-optimal: nick-to-edit distance ∈ 3–10 nt (Anzalone 2019 optimum)
    • PE3b-eligible: nicking sgRNA protospacer spans chr15:43600551 (post-edit sequence differs from pre-edit → allele-discriminating nicker)
  4. Computed GC% for each protospacer.

Deterministic; no stochastic components.

Results

Comparative matrix

VariantTotal PAMspegRNA 3–10 ntPE3b nickersBest pegRNA dist
SpCas9 NGG (baseline)5905N/A (min = 14 nt)
SpG NGN1562125.5 nt
enCas9 NGN1562125.5 nt
SpRY NRN28710223.5 nt
SpCas9-NG NG1582125.5 nt
SaCas9 NNGRRT600—

SpG / enCas9 / SpCas9-NG all return the same top hits because NGN is the intersection of their specificities and the two optimal-geometry PAMs in this window (TGA at −3.5 nt upstream, AGC at +9.5 nt downstream) are both NGN. SpRY’s extra 8 optimal pegRNAs all have NAN PAMs (CAC, AAA, GAA) and are expected to run at 40–60% of NGN activity (Walton 2020 NRN specificity landscape).

Correction to Phase 1

Phase 1 script required both protospacer_spans_variant AND nick-to-edit distance ∈ 30–100 nt. These constraints are geometrically incompatible: a protospacer that spans the variant has its nick within ~17 nt of the variant by construction (for 20-nt protospacer, nick is 17 nt from the 5’ end, so the nick can sit at most (20 − 17) = 3 nt on one side of the variant if the variant is at one edge of the protospacer, or up to ~17 nt on the other). Fixed filter drops the distance band and checks spanning only.

SpCas9 NGG PE3b nickers recovered (Phase 2):

StrandPAMNick distanceProtospacer 5′→3′GC%
−AGG12.5 ntACTGAAATTGGCACCATAGC45.0
−TGG15.5 ntGAAATTGGCACCATAGCAGG50.0
−GGG16.5 ntAAATTGGCACCATAGCAGGT45.0
+AGG11 ntCCAATTTCAGTGAAGATCTC40.0
+NGG(4th)(see JSON)—

These nickers were available from the start — Phase 1’s classic-PE3 fallback (63-nt-distant nicker TTGGTTTGGTTTCTATACCA) was unnecessary for allele discrimination.

Conservative (SpCas9 wild-type enzyme) — upgraded from Phase 1 ; superseded for lead PE3b nicker by STRC PE Phase3 Allele Discrimination:

  • pegRNA spacer: GCCCAGCTCCCCACCTGCTA (+ strand, 14 nt from edit — Phase 1 design unchanged)
  • PE3b nicker (Phase 2 lead): ACTGAAATTGGCACCATAGC (− strand, PAM AGG, 12.5 nt from edit, GC 45%) — Phase 3 shows the variant sits at protospacer position 5 (DISTAL) → weak discrimination. Replaced by Phase 3 lead CCTGAGATCTTCACTGAAAT (PAM TGG, position 17 seed, 0.5 nt nick) for stronger discrimination; trade-off is the close nick.
  • Net: ~50% geometry penalty persists; PE3b adds 1.5–3× edit efficiency (Anzalone 2019 PE3b vs PE2 data)
  • Expected OHC PE efficiency: 10–30% with Phase 3 seed nicker (revised up from Phase 2’s 8–25% estimate)

Preferred (SpG NGN nickase) — refined by STRC PE Phase3 Allele Discrimination:

  • pegRNA spacer: TGGGGGCCTGAGATCTTCAC (− strand, PAM TGA, GC 60%, nick 5.5 nt from edit)
  • PE3b nicker: TTCACTGAAATTGGCACCAT (− strand, PAM AGC, position 8 MID region, moderate discrimination, nick 9.5 nt — the balanced pick from Phase 3)
  • Geometry penalty removed; PE3b present with validated mid-region discrimination
  • Expected OHC PE efficiency: 20–40% — Phase 3 revision up from Phase 2’s 15–35% (higher confidence after verifying letter-match and mismatch-region classification)
  • Caveat: SpG has ~70% of wild-type SpCas9 activity at matched PAMs (Walton 2020); net efficiency still ahead

Overkill (SpRY NRN): no practical gain over SpG for this variant. SpRY’s advantage is only that it finds 8 extra NAN-PAM pegRNAs of questionable activity. Use only if SpG fails wet-lab validation.

Interpretation for Misha

  • Phase 1’s conclusion that “PE3 efficiency for Misha is 5–20%” was pessimistic — driven partly by missing PE3b. Corrected range with SpCas9 PE3b: 8–25%. Further upgrade with SpG PE3b: 15–35%.
  • The dominant lever is PE3b, not PAM relaxation. PE3b availability was always there — Phase 1 missed it.
  • SpG is the least-intrusive upgrade: published, characterized, available from multiple vendors (Addgene, IDT). Geometry optimum without stretching into poorly characterized NAN territory.
  • Clinical delivery unchanged: dual-AAV split-intein PE3 with either SpCas9 or SpG; both fit the same vector architecture.

Limitations

  • SpG / SpRY / SpCas9-NG activity numbers (70%, 30–50%, 60% of wild-type respectively) are published averages — real activity at THIS protospacer is unknown.
  • Off-target landscape changes with relaxed PAMs: SpG has 2–4× more off-targets than SpCas9 at matched loci (Walton 2020); SpRY is ~10× more promiscuous. Full Cas-OFFinder scan under each variant’s PAM rules is Phase 3.
  • PE3b allele discrimination requires the nicker spacer to mismatch at the variant position before edit. This is a separate design rule (check the spacer letter at the variant position equals the WT/corrected nucleotide, not the mutant). Phase 2 scans for protospacer-spanning but does not verify post-edit match direction — that is a manual check for each top candidate before lab.
  • SaCas9 gave 0/0 purely because of window size (6 PAMs total) — a wider region scan might find SaCas9 candidates further from the edit, but they’d violate the PE geometry constraint anyway.
  • The protospacer-matching-pre-edit-vs-post-edit direction for each PE3b candidate still needs per-candidate verification. For a C→A correction at chr15:43600551 (coding) = G→T on + strand, a − strand nicker whose protospacer contains T at the corresponding position = matches post-edit.

Next steps

  1. PE3b allele-discrimination check per top candidate — confirm protospacer letter at variant position matches WT/corrected, not MUT.
  2. Cas-OFFinder off-target scans for SpG at both the pegRNA spacer and PE3b nicker. Allow 0–3 mismatches, report transcript-overlapping hits.
  3. Select lead + order: synthesize pegRNA + nicker sgRNA per SpG preferred design; validate in HEK293T dual-luciferase rescue assay with STRC-E1659A minigene. $8–12 k, 6–10 weeks.
  4. Compare SpG vs SpCas9 PE3b in the rescue assay — answers whether geometry optimum (SpG) or enzyme activity (SpCas9) wins for this variant.

Replication

cd ~/STRC/models
/opt/miniconda3/bin/python3 pe_phase2_pam_expansion.py
# outputs: pe_phase2_pam_expansion.json

Files / Models

  • ~/STRC/models/pe_phase2_pam_expansion.py — 6-variant PAM scan, PE3b + pegRNA-optimum filters
  • ~/STRC/models/pe_phase2_pam_expansion.json — full candidate sets per variant

Connections

Graph View

Backlinks