CRISPR Donor Template Plasmid Design for HDR: Homology Arms, Edit Placement, and Blocking Re-cutting

Your CRISPR HDR efficiency comes back at 0.4% and you can’t tell whether the problem is the cut, the donor, or the cell. Donor template plasmid design is the variable you control most cleanly: arm lengths, edit position relative to the cut, and whether Cas9 can re-cut your repaired allele are all decisions you make at the sequence level before any cell ever sees the construct.

This post walks through the design rules for CRISPR HDR donor template plasmids specifically — not ssODN donors, which follow a different set of conventions. It covers when to pick a plasmid in the first place, how to size the homology arms by edit type, where to place the edit relative to the Cas9 cut, how to design blocking mutations that survive silent-mutation traps, and a worked N-terminal His-tag knock-in.

When a plasmid donor is the right choice

Three signals push the donor format toward a plasmid rather than a single-stranded oligo:

• Insert size above ~120 bp. Synthetic ssODNs are routinely available up to about 200 nt, but anything beyond a small tag, a SNP fix, or a loxP site is more reliably built on a plasmid backbone.
• Need for selection or a fluorescent reporter inside the insert. Antibiotic resistance cassettes, GFP, or 2A-linked markers push you above the practical ssODN length immediately.
• Stable, re-usable reagent. A plasmid donor can be miniprepped and re-sequenced; an ssODN gets re-ordered when the tube runs out and you re-validate each lot.

If your edit is a point mutation or a short tag (FLAG, single HA, single Myc), an ssODN is usually the cheaper and faster path. If the insert is a fluorescent protein, a full-length cassette, or anything with internal regulatory elements, build the donor on a plasmid.

Homology arm length by edit type

The published guidance for plasmid donors converges on two practical ranges, with a strong dependency on insert size. The numbers below are widely cited starting points from Addgene’s CRISPR 101 HDR primer and the optimized-parameter analysis from Liu et al. 2021 in Scientific Reports:

Where to put the edit relative to the Cas9 cut

This is the design rule most often missed. HDR efficiency at a given position falls off rapidly with distance from the Cas9 cut site. The practical guidance: place the edit as close to the cut site as you can — ideally within 10 bp, and within 30 bp at most. Edits placed >30 bp from the cut typically convert at less than half the rate of edits at the cut site, and by ~100 bp the efficiency is approaching zero in many published systems.

Two consequences follow:

• Your choice of sgRNA is partly constrained by where your edit needs to go. If the obvious guide cuts 80 bp away from your intended SNP, look for a different guide that cuts closer — even a guide with a slightly lower predicted efficiency score is usually a better trade than the distance penalty.
• For tag knock-ins at a start or stop codon, design the guide so it cuts within a few nucleotides of the codon you’re modifying. Then build the donor with the tag positioned at that codon, with both homology arms anchored close to the cut.

This is the same logic that drives primer design for site-directed mutagenesis: the mismatched base wants to be near the center of the homology, not at the edge.

Blocking re-cutting in your HDR donor: PAM, seed, or split

If your repaired allele still matches the sgRNA target and PAM, Cas9 will re-cut after HDR has happened. The repair pathway tries again, often producing indels and undoing your edit. Addgene’s practical writeup calls this the single biggest avoidable cause of low recovery of correctly-edited clones — adding blocking mutations can improve recovery of correctly-edited clones substantially in published comparisons.

You have three options for breaking the target site in the donor:

1. Mutate the PAM. For SpCas9, the PAM is NGG. Changing GG to anything else (NCC, NAA, NTT) abolishes recognition. This is the cleanest option when the PAM falls in a non-coding region, an intron, or a degenerate position in a coding region. If your edit lands inside the protein you’re tagging, check that the PAM-disrupting change is a silent mutation in that frame, or that it falls in a UTR.
2. Mutate the seed region. If the PAM is in a position where you can’t change it without altering the protein, change three or more nucleotides in the 8–12 bp of guide RNA target immediately adjacent to the PAM (the “seed” region). Cas9 tolerates distal mismatches but not seed mismatches. Single-base mismatches in the seed are not reliably blocking — aim for at least three changes.
3. Split the target across the insert. If your insert sits inside the guide target sequence itself, the insertion physically separates the protospacer from the PAM. Cas9 can’t re-cut a target that no longer exists as a contiguous sequence. This is the structural option and works automatically for cassette knock-ins that land in the middle of the guide.

Worked example: N-terminal His6 tag at an endogenous locus

Suppose you want to knock an N-terminal His6 tag onto an endogenous gene so the native promoter and 5′ UTR drive expression. Working backward from the design rules:

1. Pick the sgRNA. Choose a guide that cuts within ~10 bp of the start codon ATG. Score it for off-targets in your genome of interest; rule out any guide with a high-confidence off-target in a coding exon.
2. Set the insert. The insert is your tag plus a short linker: ATGCATCATCATCATCATCATGGCAGC (Met + His6 + Gly-Ser linker), 27 bp total. The endogenous ATG is replaced by this construct, with the rest of the protein in-frame downstream.
3. Homology arms. Insert is ~27 bp, well within the “short tag” range. Use 250 bp arms on each side. The left arm ends just upstream of the original ATG; the right arm begins at the codon immediately after the ATG and extends 250 bp into the coding sequence.
4. Blocking mutation. The original ATG is replaced by your tag construct, so the protospacer is already disrupted at the cut site. Check the residual sequence against the sgRNA: if there’s still a near-perfect match plus an intact PAM downstream of the insertion, introduce 2–3 silent mutations in the first few codons of the right arm to break the seed region.
5. Assemble the donor. Synthesize the full 527 bp cassette (250 + 27 + 250) as a gBlock or PCR-assemble it, then clone into a high-copy backbone (pUC19, pBluescript) for miniprep yields and sequence verification.
6. Verify the donor before transfection. Sequence the full insert + both junctions. A common failure mode is a frame-shift introduced during cloning — the construct looks right on a restriction map but the open reading frame is broken.

Verifying integration, not just presence

After transfection and selection, most clones that screen positive by PCR are random integrants, not HDR-edited cells. The PCR primer pair you use to screen has to distinguish on-target HDR from random integration of the donor plasmid.

The diagnostic primer design: one primer outside the homology arm (in genomic sequence not present on the donor), one primer inside the insert. This pair produces a band only when the insert has integrated at the correct locus. Verify the primer Tms match using a Tm calculator with the same nearest-neighbor model for both primers, and design the genomic primer at least 50–100 bp outside the homology arm to give the diagnostic specificity. The same colony PCR practice that we covered for verifying insert orientation after cloning applies here: design the primers to give a band of unambiguous size, and run a known-positive control.

A second-line check is a diagnostic restriction digest of a PCR product spanning both junctions, or Sanger sequencing across both junctions from a PCR product amplified with two outside-the-arm primers. If you’ve introduced silent mutations as blocking mutations, sequencing will read those as confirmation of correct HDR. Use the restriction enzyme calculator to find an enzyme that cuts differently in the HDR allele versus the wild-type or random-integrant alleles.

Where to escalate

For most knock-in projects, plasmid donor design follows the rules above and the experimental variable that dominates outcome is cell-cycle synchronization, Cas9 delivery format (RNP versus plasmid), and clonal screening throughput. If you’ve verified that your donor follows the arm-length and blocking-mutation rules and HDR efficiency is still low, the next levers are biological: timing the edit during S/G2, using a Cas9 fused to a DNA repair domain, or switching to ssODN if your insert can be shortened to fit. Donor design is the cleanest variable to control, but it’s not the only one that matters.