# CRISPR interference
CRISPR interference (CRISPRi) is a gene repression technique in which a catalytically dead Cas protein, guided by a small guide RNA, binds a chosen DNA sequence and blocks transcription of the target gene without cutting or otherwise altering the DNA. The method was introduced in 2013 by [Lei S. Qi](https://www.edgepedia.org/lei-stanley-qi) and colleagues, who repurposed the CRISPR-associated protein Cas9 as a programmable transcriptional blocker in bacteria.[1](https://doi.org/10.1016/j.cell.2013.02.022) The underlying enzyme, dCas9, comes from work by [Martin Jinek](https://www.edgepedia.org/martin-jinek) and colleagues showing that Cas9 is a dual-RNA-guided DNA endonuclease.[2](https://doi.org/10.1126/science.1225829) CRISPRi reduces mRNA and, downstream, protein levels; it is mechanistically distinct from [RNA interference](https://www.edgepedia.org/rna-interference), which destroys already-transcribed mRNAs, and from cutting-based CRISPR knockout, which edits the genome.[1](https://doi.org/10.1016/j.cell.2013.02.022) With effector-domain fusions, repression typically reaches 90–99% knockdown, making CRISPRi a standard tool for functional genomics screens.[3](https://doi.org/10.1016/j.cell.2014.09.029)
| Key fact | Detail |
|---|---|
| Effector protein | dCas9, S. pyogenes Cas9 with D10A (RuvC) and H840A (HNH) mutations, binds DNA but cannot cleave it[4](https://pmc.ncbi.nlm.nih.gov/articles/PMC3922765/) |
| Output | Reduced mRNA and protein; no change to the target DNA sequence[1](https://doi.org/10.1016/j.cell.2013.02.022) |
| Bacterial repression | Up to 1,000-fold in E. coli; nontemplate-strand guides give 10- to 300-fold[1](https://doi.org/10.1016/j.cell.2013.02.022) |
| Mammalian repression | dCas9-KRAB typically 90–99% knockdown; dCas9 alone is modest (46% of an EGFP reporter in HEK293)[3](https://doi.org/10.1016/j.cell.2014.09.029) • [1](https://doi.org/10.1016/j.cell.2013.02.022) |
| Target region (mammalian) | −50 to +300 bp around the TSS, peak ~50–100 bp downstream; later work recommends +25 to +75 bp[3](https://doi.org/10.1016/j.cell.2014.09.029) • [5](https://doi.org/10.1038/s41467-018-07901-8) |
| Reversibility | Conventional CRISPRi repression is reversible and inducible and does not alter the DNA sequence; CRISPRoff-like epigenetic variants can establish durable, heritable silencing[3](https://doi.org/10.1016/j.cell.2014.09.029) |
| Main uses | Genome-wide loss-of-function screens, essential-gene studies, synthetic circuits, therapeutic repression[6](https://www.nature.com/articles/s41467-018-04899-x) |
## How it works
dCas9 is Cas9 with two point mutations, D10A in the RuvC-like domain and H840A in the HNH domain, which abolish nuclease activity while leaving DNA binding intact.[4](https://pmc.ncbi.nlm.nih.gov/articles/PMC3922765/) A single guide RNA (sgRNA) with a 20-nucleotide complementary region directs dCas9 to any sequence next to an NGG PAM. Because the protein cannot cut, the bound dCas9:sgRNA complex acts as a roadblock: repression arises from physical collision between the elongating [RNA polymerase](https://www.edgepedia.org/rna-polymerase) and the dCas9:sgRNA complex, and can also interfere with RNA polymerase binding or transcription-factor binding.[1](https://doi.org/10.1016/j.cell.2013.02.022)
Position and strand determine strength. In E. coli, sgRNAs targeting the nontemplate strand of the coding region gave 10- to 300-fold repression, whereas template-strand guides had little effect; promoter targeting at the −35 box gave roughly 100-fold repression, and repression declines with distance from the transcription start site (TSS).[1](https://doi.org/10.1016/j.cell.2013.02.022) NET-seq confirmed that nontemplate-strand binding blocks elongation.[4](https://pmc.ncbi.nlm.nih.gov/articles/PMC3922765/) In mammalian cells, dCas9 alone only modestly blocks transcription, so repression there relies on fused repressor domains such as KRAB, which recruits KAP1/TRIM28, SETDB1, HP1, and DNMTs to establish repressive chromatin.[7](https://doi.org/10.1016/j.cell.2013.06.044) • [8](https://www.mdpi.com/1422-0067/27/8/3564)
## How it is done
A CRISPRi experiment has four practical decisions: guide design, target region, dCas9 delivery, and readout.
**Guide design.** The sgRNA consists of a 20-nt target-specific spacer followed by an approximately 80-nt Cas9-binding scaffold derived from the crRNA repeat and tracrRNA; scaffold designs vary, and a separate Pol III termination signal (typically a short poly-U tail) is added by the expression system.[4](https://pmc.ncbi.nlm.nih.gov/articles/PMC3922765/) Mismatches in the 12-nt PAM-adjacent seed reduce repression by 70–90%, while truncating the base-pairing region below 20 nt substantially dampens repression.[4](https://pmc.ncbi.nlm.nih.gov/articles/PMC3922765/) A 14-nt specificity check (12-nt seed plus 2 PAM nucleotides) is recommended, discarding guides with more than one genomic binding site.[4](https://pmc.ncbi.nlm.nih.gov/articles/PMC3922765/)
**Target region.** In mammalian cells, guides should fall in a window of −50 to +300 bp relative to the TSS, searching for GN(19)NGG motifs; Gilbert et al. found peak activity ~50–100 bp downstream of the TSS, while the crisprDesign tutorial, citing Sanson et al. (2018), recommends the +25 to +75 bp region for optimal inhibition.[9](https://cshprotocols.cshlp.org/content/2016/1/pdb.prot090175.full) • [3](https://doi.org/10.1016/j.cell.2014.09.029) • [10](https://github.com/crisprVerse/Tutorials/tree/master/Design_CRISPRi) In bacteria, target the nontemplate strand of the 5′ coding region or 5′-UTR, or the −35/−10 promoter boxes.[4](https://pmc.ncbi.nlm.nih.gov/articles/PMC3922765/)
**Delivery and tuning.** The original bacterial system used two plasmids: anhydrotetracycline (aTc)-inducible dCas9 on p15A and a constitutive sgRNA on ColE1.[11](https://www.addgene.org/browse/article/6441/) In mammalian cells, dCas9-KRAB is commonly delivered on plasmids or lentivirus, or knocked into the AAVS1 safe-harbor locus under doxycycline control for inducible, multi-gene silencing in human pluripotent stem cells.[12](https://pubmed.ncbi.nlm.nih.gov/39083383/) The dCas9 cassette often exceeds the 4.7 kb packaging limit of a single AAV vector, motivating lipid nanoparticle (LNP) delivery.[8](https://www.mdpi.com/1422-0067/27/8/3564) Leaky expression from inducible promoters can cause ~80% background repression, so low-copy plasmids or weaker promoters are used for tunable experiments; repression is achievable within 1–2 weeks.[4](https://pmc.ncbi.nlm.nih.gov/articles/PMC3922765/) Knockdown is quantified by qPCR using the \( 2^{-\Delta\Delta C_{t}} \) method, with \( \Delta\Delta C_{t} = \Delta C_{t}(\text{CRISPRi}) - \Delta C_{t}(\text{control}) \).[9](https://cshprotocols.cshlp.org/content/2016/1/pdb.prot090175.full)
## Origin
The foundation was the 2012 report by Jinek, Chylinski, Fonfara, Hauer, Doudna, and Charpentier that Cas9 is a programmable dual-RNA-guided endonuclease, which established the guide-RNA design used throughout CRISPRi.[2](https://doi.org/10.1126/science.1225829) CRISPRi itself was introduced in 2013 by Qi, Larson, Gilbert, Doudna, Weissman, Arkin, and Lim in Cell, showing dCas9 plus sgRNA represses transcription in E. coli with up to 1,000-fold repression and no detectable off-target effects.[1](https://doi.org/10.1016/j.cell.2013.02.022) The same year, [Luke A. Gilbert](https://www.edgepedia.org/luke-a-gilbert) and colleagues extended the approach to eukaryotes by fusing dCas9 to effector domains, enabling stable repression and activation in human and yeast cells.[7](https://doi.org/10.1016/j.cell.2013.06.044) In 2014, Gilbert and colleagues built genome-scale CRISPRi and CRISPRa libraries and defined the TSS-targeting rules.[3](https://doi.org/10.1016/j.cell.2014.09.029) Earlier repression precursors were RNA interference, zinc-finger proteins, and TALE proteins, though zinc-finger and TALE constructs are time-consuming and expensive to engineer for new targets.[4](https://pmc.ncbi.nlm.nih.gov/articles/PMC3922765/)
## Variants
**Effector fusions.** dCas9-KRAB repressed a GFP reporter 5-fold versus 2-fold for dCas9 alone, and stable dCas9-KRAB silenced endogenous human genes 5- to 15-fold and 50-fold in yeast.[7](https://doi.org/10.1016/j.cell.2013.06.044) Adding MeCP2 produced an enhanced repressor (Nan Cher Yeo and colleagues, [Nature Methods](https://www.edgepedia.org/nature-methods) 2018).[13](https://doi.org/10.1038/s41592-018-0048-5) Most platforms use the KRAB domain from KOX1 (ZNF10); alternative KRAB domains, particularly ZIM3, improve silencing, and dCas9-ZIM3(KRAB)-MeCP2(t) is a next-generation tripartite repressor.[14](https://link.springer.com/article/10.1186/s13059-025-03640-4)
**Recruitment and multiplexing.** Scaffold RNAs (scRNAs) recruit different effectors to different genes, allowing simultaneous activation and repression.[9](https://cshprotocols.cshlp.org/content/2016/1/pdb.prot090175.full) On the Cas12a side, a hyper-efficient dHyperLbCas12a repressor supports pre-crRNA arrays with up to 14 crRNAs from a single [RNA polymerase II](https://www.edgepedia.org/rna-polymerase-ii) promoter, and hybrid arrays can activate and repress genes in the same cell.[15](https://www.nature.com/articles/s41467-026-69090-z) The CRISPRoff system, a KRAB-dCas9-DNMT3A-DNMT3L fusion, converts transient repression into long-term, heritable silencing.[16](https://www.mdpi.com/2673-8856/5/4/57)
## Applications
**Genome-wide screens.** A genome-scale E. coli CRISPRi library of 55,671 sgRNAs covered 98.6% of 4,140 protein-coding genes, exceeding [Tn-seq](https://www.edgepedia.org/tn-seq) coverage, and against the Keio essential-gene standard reached AUC-ROC 0.952.[6](https://www.nature.com/articles/s41467-018-04899-x) In mammalian cells, genome-scale libraries of 10 sgRNAs per gene were validated in pooled screens, with 99.7% of negative-control sgRNAs showing no detectable activity and qPCR confirming ~80–99% knockdown per sgRNA.[3](https://doi.org/10.1016/j.cell.2014.09.029)
**Essential genes and tunable control.** Because repression strength can be tuned by guide-target mismatches rather than dCas9 concentration, and tuned repression adds no extrinsic noise, CRISPRi suits partial knockdown of essential genes in synthetic circuits.[17](https://link.springer.com/article/10.15252/msb.20177899) In human iPSCs, CRISPRi induces specific and reversible gene silencing (Mohammad A. Mandegar and colleagues, 2016), and a 2024 doxycycline-inducible protocol achieved more than 96–99% knockdown of all FOXA genes during endoderm differentiation.[18](https://doi.org/10.1016/j.stem.2016.01.022) • [12](https://pubmed.ncbi.nlm.nih.gov/39083383/)
**Therapeutics.** CRISPRi-derived epigenetic silencing therapies have entered clinical testing: the AAVrh74-delivered epigenetic editor EPI-321 is in a Phase 1/2 first-in-human trial for facioscapulohumeral muscular dystrophy (NCT06907875)[8](https://www.mdpi.com/1422-0067/27/8/3564), and the investigational LNP-delivered dCas9-KRAB-DNMT3A/3L silencer CRMA-1001 is in development for chronic hepatitis B[16](https://www.mdpi.com/2673-8856/5/4/57); a single dose of LNP-encapsulated CRISPRi silenced the target oncogene in mice for at least one year.[16](https://www.mdpi.com/2673-8856/5/4/57) • [8](https://www.mdpi.com/1422-0067/27/8/3564)
## Limitations and alternatives
**Efficiency and position dependence.** Only 40% of TSS-proximal sgRNAs were effective before optimization; a support-vector-machine model of target-TSS distance using CAGE-seq annotation raised the effective fraction to 70%.[19](https://dash.harvard.edu/bitstreams/7312037e-3856-6bd4-e053-0100007fdf3b/download) In bacteria, active guides cluster in the first 5% of the open reading frame, and the NGG PAM requirement makes short genes and extreme-GC genomes hard to target.[6](https://www.nature.com/articles/s41467-018-04899-x)
**Failure modes.** CRISPRi can show incomplete knockdown, variability across cell lines and targets, and guide-sequence-dependent inconsistency.[14](https://link.springer.com/article/10.1186/s13059-025-03640-4) Off-target dCas9 binding is highly variable: 10 to 1,300 off-target binding sites per sgRNA were observed in a screen of 12 sgRNAs, with other studies reporting up to 6,000, mostly in open chromatin.[16](https://www.mdpi.com/2673-8856/5/4/57) In operons, repression of an upstream gene has polar effects on downstream genes, including a position-dependent reverse-polarity effect within the first 50 bp of the next gene.[6](https://www.nature.com/articles/s41467-018-04899-x) Bidirectional promoters, present in up to 10% of human genes, score as false positives in CRISPRi screens.[19](https://dash.harvard.edu/bitstreams/7312037e-3856-6bd4-e053-0100007fdf3b/download) Overexpressed dCas9 has been reported to be toxic for E. coli; a molecular-glue system that degrades Cas9 in the presence of the FDA-approved drug pomalidomide offers a control mechanism.[17](https://link.springer.com/article/10.15252/msb.20177899) • [8](https://www.mdpi.com/1422-0067/27/8/3564)
**Comparison with alternatives.** Against nuclease-active CRISPR knockout (CRISPRc), cutting-based screens identified 98% of previously defined essential genes, while TSS-optimized CRISPRi identified 92% of core essential genes; CRISPRi avoids the copy-number-amplification bias of cutting screens.[19](https://dash.harvard.edu/bitstreams/7312037e-3856-6bd4-e053-0100007fdf3b/download) Knockouts are irreversible and can generate in-frame-indel subpopulations, whereas CRISPRi induces no DNA damage or DNA-repair activation and its repression is reversible.[14](https://link.springer.com/article/10.1186/s13059-025-03640-4) CRISPRa, which targets activator fusions to a −400 to −50 bp window upstream of the TSS, is the complementary gain-of-function counterpart, and CRISPRi/a together modulate expression over a ~1,000-fold range.[9](https://cshprotocols.cshlp.org/content/2016/1/pdb.prot090175.full) • [3](https://doi.org/10.1016/j.cell.2014.09.029) Compared with RNAi, CRISPRi acts at the DNA level to block transcription rather than destroying mRNAs.[1](https://doi.org/10.1016/j.cell.2013.02.022)
## References
1. [Lei S. Qi and colleagues (2013). Repurposing CRISPR as an RNA-Guided Platform for Sequence-Specific Control of Gene Expression. Cell.](https://doi.org/10.1016/j.cell.2013.02.022)
2. [Martin Jinek and colleagues (2012). A Programmable Dual-RNA–Guided DNA Endonuclease in Adaptive Bacterial Immunity. Science.](https://doi.org/10.1126/science.1225829)
3. [Luke A. Gilbert and colleagues (2014). Genome-Scale CRISPR-Mediated Control of Gene Repression and Activation. Cell.](https://doi.org/10.1016/j.cell.2014.09.029)
4. [CRISPR interference (CRISPRi) for sequence-specific control of gene expression (Qi et al., Nature Protocols 2013; PMC full text)](https://pmc.ncbi.nlm.nih.gov/articles/PMC3922765/)
5. [Kendall R. Sanson and colleagues (2018). Optimized libraries for CRISPR-Cas9 genetic screens with multiple modalities. Nature Communications.](https://doi.org/10.1038/s41467-018-07901-8)
6. [Pooled CRISPR interference screening enables genome-scale functional genomics study in bacteria with superior performance (Peters et al., Nature Communications 2018)](https://www.nature.com/articles/s41467-018-04899-x)
7. [Luke A. Gilbert and colleagues (2013). CRISPR-Mediated Modular RNA-Guided Regulation of Transcription in Eukaryotes. Cell.](https://doi.org/10.1016/j.cell.2013.06.044)
8. [CRISPR Interference to Inhibit Oncogenes for Cancer Therapy (Int. J. Mol. Sci. review)](https://www.mdpi.com/1422-0067/27/8/3564)
9. [CRISPR Technology for Genome Activation and Repression in Mammalian Cells (CSH Protocols 2016)](https://cshprotocols.cshlp.org/content/2016/1/pdb.prot090175.full)
10. [gRNA design for CRISPR interference (crisprVerse tutorial)](https://github.com/crisprVerse/Tutorials/tree/master/Design_CRISPRi)
11. [Addgene: Repurposing CRISPR as an RNA-Guided Platform for Sequence-Specific Control of Gene Expression (author summary)](https://www.addgene.org/browse/article/6441/)
12. [Protocol for establishing inducible CRISPR interference system for multiple-gene silencing in human pluripotent stem cells (STAR Protocols 2024; PubMed record)](https://pubmed.ncbi.nlm.nih.gov/39083383/)
13. [Nan Cher Yeo and colleagues (2018). An enhanced CRISPR repressor for targeted mammalian gene regulation. Nature Methods.](https://doi.org/10.1038/s41592-018-0048-5)
14. [Engineering novel CRISPRi repressors for highly efficient mammalian gene regulation (Genome Biology 2025)](https://link.springer.com/article/10.1186/s13059-025-03640-4)
15. [dHyperCas12a enables multiplexed CRISPRi screens (Nature Communications 2026)](https://www.nature.com/articles/s41467-026-69090-z)
16. [CRISPR-Based Transcriptional Regulation: Technologies, Applications, and Future Directions (review, 2024/2025)](https://www.mdpi.com/2673-8856/5/4/57)
17. [Tuning dCas9's ability to block transcription enables robust, noiseless knockdown of bacterial genes (Molecular Systems Biology 2018)](https://link.springer.com/article/10.15252/msb.20177899)
18. [Mohammad A. Mandegar and colleagues (2016). CRISPR Interference Efficiently Induces Specific and Reversible Gene Silencing in Human iPSCs. Cell stem cell.](https://doi.org/10.1016/j.stem.2016.01.022)
19. [Comparing CRISPRc and CRISPRi loss-of-function screens (deposited manuscript)](https://dash.harvard.edu/bitstreams/7312037e-3856-6bd4-e053-0100007fdf3b/download)
---
*Topic: Encyclopedia › Life and health › Biological foundations › RNA and gene regulation › Small regulatory RNAs*
*Initially written Sep 29, 2026 · Edited: Oct 15, 2026*
*Copyright 2026 EdgeChat AI, a subsidiary of Biostate AI.*
License: Edgepedia Community License 1.0, https://www.edgepedia.org/edgepedia/license