Expressing Many sgRNAs From One Plasmid
6 min read · Updated October 30, 2026
Multiplexed CRISPR runs into a problem that has nothing to do with the guides. An sgRNA scaffold is 60 to 80 bases of fixed sequence — 61 for the handles this tool draws on — so expressing twenty guides from one plasmid the obvious way puts twenty identical copies of it into one molecule, along with twenty copies of the same promoter and terminator.
That construct recombines with itself in the cell and synthesis vendors refuse it. The fix is not a cleverer layout; it is a different set of parts.
Why the scaffold cannot simply be randomised
The obvious answer to twenty identical scaffolds is to vary them. The obvious answer is wrong: the scaffold's secondary structure is what Cas9 binds, so changing bases at random destroys the part while leaving the sequence looking fine.
What is needed is a set of scaffold variants that are functionally equivalent and sequence-divergent, and that cannot be derived — it has to be measured. Such a collection exists, published with the method that produced it: twenty-seven scaffold variants, plus promoters and neutral spacers selected the same way.
So the capacity of this approach is a property of a parts collection rather than of your design, and the first number to look at is how many parts that collection offers at the repeat threshold you want.
What a repeat threshold costs in capacity
The threshold is the longest stretch any two parts in the finished array may share. Lower is a more stable construct built from fewer available parts, and the trade is steep in exactly one of the three pools.
At a 12 bp threshold the collection offers 28 of its 29 promoters, 12 of its 27 scaffold handles and all 62 spacers — so twelve transcription units fit. At 15 bp: 29 promoters, 18 handles, 62 spacers, and eighteen units fit. At 20 bp: 24 handles and twenty-four units.
The handle pool is the constraint at every threshold, and 27 is the size of that pool — the most that actually assembles is 26, at the loosest threshold the tool allows. An array of forty guides is not a design problem; it is outside what the parts support.
Selection has to be across the pools, not within them
The subtle failure is to pick non-repetitive promoters from the promoter pool, non-repetitive handles from the handle pool, and non-repetitive spacers from the spacer pool, and then put them together. A promoter and a handle can each be clean within their own pool and share eighteen bases with each other, which leaves the array exactly as repetitive as no selection at all.
So the selection runs across all three part types at once. The cost is visible in the capacity numbers above: the limit is set by the pools jointly, not by the smallest pool on its own.
The finished molecule has to be re-measured
A repeat spanning the end of one handle and the start of the next promoter exists in neither part. It is created by the junction, which means no amount of care about the parts catches it, and it is only visible once the array is assembled.
A five-guide array built from this collection at a 12 bp threshold comes out at 962 bp — five transcription units of 149 to 156 bp, separated by four insulating spacers, drawing 5 promoters, 5 handles and 4 spacers from the pools — and is then re-scanned as one molecule, which found no repeat over the threshold. That scan is the check that matters, and a design that passes part-wise and fails here is a design that would have been ordered.
What this does not do
It takes guides; it does not choose them. Which 20-mer to target, whether it cuts efficiently, and what else in the genome it matches are three separate questions with their own tools, and none of them is answered by the array layout.
Nor does it predict expression. Drawing each unit's promoter from a non-repetitive set means the units differ in strength, and the array tells you which promoter each unit got rather than what that implies for the guide's abundance.
Frequently asked questions
Why can I not just repeat the same sgRNA scaffold twenty times?
Because twenty identical 61-base stretches in one molecule recombine. The cell deletes between the copies — losing guides silently, so the screen returns a subset working — and synthesis vendors flag the repetition before the molecule exists. The construct is not wrong in any way a sequence check notices.
How many sgRNAs can one array express?
Fewer than the 27 scaffold variants published, because every unit needs a handle no other unit uses and the selection runs across the promoter and spacer pools at the same time. In practice: 26 units at the loosest threshold the tool allows, 24 at 20 bp, 18 at 15 bp and 12 at 12 bp. A request for all 27 is refused, naming how many handles the collection could supply.
Why does a lower repeat threshold give me fewer guides?
Because the threshold decides which parts count as mutually non-repetitive, and tightening it disqualifies parts. The scaffold handle pool is the one that shrinks fastest — 12 of 27 at a 12 bp threshold against 24 of 27 at 20 bp — so the array capacity follows it.
Can I design my own scaffold variants?
Not by randomising. The scaffold's structure is what Cas9 binds, so sequence changes have to be shown not to break it, which means measuring them rather than generating them. That is why these collections are published as data.
Related references
Related tools
Build a multiplexed CRISPR array that expresses up to 27 sgRNAs from one construct and shares no long stretch with itself — the property that stops it recombining away and getting refused by synthesis vendors.
Scan a sequence for SpCas9, SaCas9 or Cas12a guide candidates with PAMs and scoring.
Find the largest subset of your parts that share no long stretch of sequence, or design new ones that do not — the property that keeps a multi-part assembly stable and synthesisable.
Measure the repeats, GC extremes, GC transitions, homopolymer runs and hairpins that make a fragment hard to synthesise — each against the threshold vendors publish, with no score and no probability.
Build a multi-gene operon from a promoter, coding sequences and a terminator, then scan the assembled molecule for the internal promoters, ribosome binding sites and terminators that the joins created.