CRISPR Constructs

How CRISPR-Cas9 Works

CRISPR-Cas9 is a genome editing system adapted from bacterial immune defense. It consists of two components: the Cas9 protein, which acts as molecular scissors, and a guide RNA (gRNA), which directs Cas9 to a specific location in the genome. The guide RNA is a short sequence — typically 20 nucleotides — complementary to the target DNA. When the guide RNA binds its target, Cas9 creates a double-strand break at that precise location.

The cell’s DNA repair machinery then fixes the break. If no template is provided, the repair is error-prone (non-homologous end joining, or NHEJ), typically introducing small insertions or deletions that disrupt the gene — a knockout. If a repair template is provided (homology-directed repair, or HDR), the cell incorporates the template sequence, enabling precise edits such as point mutations or gene insertions. Scoville Splice uses both strategies: NHEJ for the POX knockout and HDR for overexpression cassettes and Pun1 point mutations.

Guide RNA Design

Each of the 24 guide RNAs was computationally designed to maximize on-target efficiency while minimizing off-target activity. The design process considers several factors:

• PAM site selection. Cas9 requires a protospacer adjacent motif (PAM) — the sequence NGG — immediately downstream of the target site. Guide RNAs are designed only at positions where a suitable PAM exists.
• On-target scoring. Computational models predict how efficiently a given guide RNA will direct Cas9 cutting. Guides with high predicted activity scores are prioritized.
• Off-target minimization. Each candidate guide is aligned against the entire Capsicum genome to identify potential off-target binding sites. Guides with close matches elsewhere in the genome are rejected or deprioritized.
• GC content. Guide RNAs with extreme GC content (very high or very low) tend to have reduced activity. The 24 selected guides fall within the optimal 40-70% GC range.

The Tiered Construct Architecture

The construct architecture is organized into tiers, with each successive tier adding one additional construct to the previous tier’s configuration. This modular approach is guided by metabolic flux analysis: the construct with the highest impact on the current bottleneck is added first, and each subsequent construct addresses the next limiting factor.

Computational Design First

Every construct in the Scoville Splice architecture was computationally designed, modeled, and validated in silico before any wet-lab synthesis. Guide RNAs were screened against the full Capsicum genome for off-target activity. Overexpression cassettes were codon-optimized for the target cultivar. Pun1 mutations were scored using ESM2’s protein language model. Metabolic flux simulations predicted the SHU outcome of each construct combination before a single plant was transformed.

This computational-first approach dramatically reduces the number of physical experiments required. Rather than screening thousands of random mutations, the design space is narrowed to high-confidence candidates before any DNA is synthesized. The 25 constructs represent the computationally optimal solution, not a sample from a larger library.

Public Record

The complete construct sequences have been deposited under GenBank accession SUB16548149. This public submission provides the research community with full access to the designed sequences and the tiered construct architecture used across the Scoville Splice platform.

By the Numbers