Inducible expression of large gRNA arrays for multiplexed CRISPRai applications

Cells exhibit complex behaviors through regulated gene expression. Tools that enable targeted activation and repression of endogenous genes allow reprogramming of cells for basic research and biotechnological applications. CRISPRa and CRISPRi have emerged as powerful approaches for modulating gene expression, and their coordinated use (CRISPRai) enables simultaneous up- and down-regulation to explore transcriptional landscapes and redirect metabolic fluxes. In standard CRISPRai implementations, a single dCas protein (often dCas9) is guided by gRNAs to promoters or coding regions to activate or inhibit transcription, but effective inhibition often requires a repression domain and can be limited by available PAM sites. More effective CRISPRai uses orthogonal Cas proteins (e.g., dCas9 and dCas12a) each fused to activation or repression domains, or a single Cas with modified gRNAs recruiting activators or repressors via orthogonal aptamers. However, mixed-identity gRNAs and complex cloning reduce scalability, limiting multiplexing (e.g., only up to four gRNAs in S. cerevisiae in prior CRISPRai studies). Although methods exist for assembling polycistronic gRNA arrays in various organisms, combined activation and repression from large arrays had not been realized. Inducibility is also desirable to mitigate fitness costs, enable manipulation of essential genes, and improve genetic stability by allowing growth prior to activation. Previous work showed inducible large arrays only for CRISPRa or CRISPRi separately. This study addresses these gaps by developing a near leak-free, inducible, large polycistronic gRNA array system enabling simultaneous CRISPRa and CRISPRi in S. cerevisiae, with high multiplexing capacity and tight inducible control.

Prior CRISPR-based transcriptional regulation strategies include dCas9-mediated activation and inhibition, with inhibition enhanced by fusion to repression domains. Orthogonal Cas systems (dCas9 and dCas12a) have enabled coordinated activation and repression on separate targets, and modified gRNA scaffolds with protein-binding aptamers have recruited activators and repressors to a single Cas. Despite these advances, mixed-identity gRNAs complicate cloning and limit multiplexing: earlier S. cerevisiae CRISPRai efforts expressed up to four gRNAs. Advanced assembly methods have enabled polycistronic arrays for multiple gRNAs in other contexts (e.g., up to 7 mixed-identity gRNAs in E. coli, 12 Cas9 gRNAs in yeast, 25 Cas12a gRNAs in mammalian cells), but had not been applied to simultaneous activation and repression. Inducible CRISPR regulation has been demonstrated for large arrays in either CRISPRa or CRISPRi, but not for combined CRISPRai multiplexing. These limitations motivated a system that achieves high multiplexing with inducible, simultaneous activation and repression.

The authors engineered an inducible, multiplexed CRISPRai system centered on polycistronic gRNA arrays processed by Csy4. Key steps and components: 1) Discovery and challenge: When initially implementing Tet-inducible expression for long gRNA arrays (Csy4-processed, Pol II-driven) for CRISPRi, the team observed substantial leakiness even with low-leak and leak-free promoters. They discovered that long gRNA arrays targeting promoter regions can drive their own transcription even without an upstream promoter, likely due to promoter-derived sequences within guides clearing nucleosomes and enabling transcription initiation. 2) Inducible array silencing design (Design 3): To ensure near leak-free control, they implemented opposing Tet-ON and Tet-OFF systems. Tet-ON: rtTA-Gal4 binds TetO sites upstream of a minimal core promoter in the presence of anhydrotetracycline (aTc) to drive array transcription. Tet-OFF: a mutated TetR (E37A P39K) fused to the Mxi1 repression domain (mutTetR-Mxi1) binds an orthogonal operator (mutTetO, Tet4C5G) distributed throughout the array (interspersed between gRNA clusters) to silence transcription across the full array in the absence of aTc. Upstream sites target a minimal core promoter adapted from Chen et al. 3) Performance characterization of inducible arrays: Arrays targeting fluorescent reporters (18 gRNAs against ALD6, TEF1, HHF1 promoters driving mScarlet-I, mGFPmut2, mTagBFP2) were tested across promoter designs. Design 3 yielded 96–98% of maximum reporter expression in the uninduced state, strong induction upon 1 µM aTc (no significant difference from constitutive expression), and up to 111-fold dynamic range. Silencing remained tight with up to six gRNAs between mutTetO sites. 4) CRISPRai protein architecture and tuning: The system uses orthogonal dCas12a-VP for activation and dCas9-Mxi1 for repression, guided by their respective gRNAs from a single Csy4-processed array. Expression levels of dCas12a-VP, dCas9-Mxi1, and Csy4 were tuned using low and medium strength yeast promoters (from the Yeast MoClo Toolkit) to balance performance and growth. Reporter assays (RNR2→mRuby2 activation by dCas12a-VP gRNAs; TEF1→Venus repression by dCas9-Mxi1 gRNAs) showed minimal effect of higher protein expression on performance (no change to maximum repression, 10–20% effect on maximum activation), but higher expression reduced growth rates. Final toolkit uses weak REVI for dCas12a-VP, weak PSP2 for dCas9-Mxi1, and medium HTB2 for Csy4; rtTA-Gal4 and mutTetR-Mxi1 are driven by weak RAD27 and POP6 promoters. 5) Vector and cloning: An all-in-one genomic integration vector (HO locus) carries dCas12a-VP, dCas9-Mxi1, Csy4, rtTA-Gal4, mutTetR-Mxi1, and an inducible gRNA array cassette with interspersed mutTetO sites. The vector is available with 10 selectable markers (URA3, LEU2, HIS3, TRP1, LYS2, MET17, KanR, NatR, HygR, ZeoR). gRNA arrays are assembled by Golden Gate cloning: up to 6 gRNAs directly into the vector in one BsaI reaction; up to 24 gRNAs assembled via four sub-array plasmids and a second BsmBI step. Csy4 sites flank each guide scarlessly. BioBrick-compatible prefixes/suffixes facilitate array swapping and concatenation. 6) Experimental workflows: Reporter assays used flow cytometry to quantify fluorescent outputs in induced and uninduced states and across time courses after single induction. Growth impacts of CRISPR protein expression were measured via microplate growth curves. For metabolic engineering, an 11-guide array (9 repression gRNAs: ADH1, ADH3, FUM1, IDP1, SDH1, SDH3, SER3, SDH2, SER33; 2 activation gRNAs: ADR1, ICL1) was designed using Benchling (activation targets −200 to −350 bp, repression targets −100 to +150 bp relative to start codons). RT-qPCR assessed target transcriptional changes; succinic acid production was quantified after 2 days in SD medium via UPLC-MS, comparing wild-type, untargeted control array, and targeted array strains, with and without 1 µM aTc.

• Long gRNA arrays can self-transcribe without an upstream promoter, causing leakiness in conventional inducible designs. - Opposing Tet-ON (rtTA-Gal4) and Tet-OFF (mutTetR-Mxi1 on mutTetO sites) controlling the gRNA array yields near leak-free inducibility: 96–98% of maximum reporter expression retained in the uninduced state; no significant difference between induced arrays and constitutive arrays after 1 µM aTc induction; up to 111-fold change in reporter expression upon induction. - Effective silencing observed with up to six gRNAs between mutTetO sites; basal activity increases slightly beyond this spacing. Inducible arrays reduced transformation-associated growth defects relative to constitutive arrays. - CRISPR protein expression tuning: Increasing dCas12a-VP, dCas9-Mxi1, and Csy4 expression did not improve maximum repression and only modestly improved activation (10–20%), but reduced growth rates. Final chosen promoters (weak REVI for dCas12a-VP, weak PSP2 for dCas9-Mxi1, medium HTB2 for Csy4; rtTA-Gal4 under RAD27, mutTetR-Mxi1 under POP6) balanced performance and fitness. - Stability and dynamics: Single aTc induction produced sustained effects over at least five days; in a 3-activation/3-repression reporter array, activation of mScarlet-I increased by ~800%, and repression of mTagBFP2 decreased expression by ~90% at 24 h, remaining stable over multiple days. Arrays remained stable in the uninduced state over at least a week of passaging. - Metabolic engineering application: An inducible 11-gRNA CRISPRai array targeting central metabolism in S. cerevisiae increased succinic acid production 45-fold versus WT after 2 days (WT induced: 9.38 ± 5.7 mg/L; Targeted induced: 426.9 ± 13.3 mg/L). Induction produced a 16-fold increase relative to the uninduced targeted strain (26.4 ± 0.5 mg/L). No significant changes were observed in control conditions (WT and untargeted arrays ± aTc), indicating the increase is exclusively due to CRISPRai and is tightly inducible. - Multiplexing capacity: Toolkit supports up to 24 gRNAs (via four sub-arrays), combining activation and repression gRNAs in any order with scarless Csy4 processing.

The work addresses the need for a scalable, inducible CRISPRai platform capable of simultaneous activation and repression across many targets. By demonstrating that long gRNA arrays can self-transcribe and then solving this with distributed Tet-OFF silencing coupled to Tet-ON induction, the authors achieve tight control over array transcription. This enables near leak-free operation, reduced burden before induction, and improved strain stability—critical when targeting growth-related or essential genes. Tuning CRISPR protein expression minimized fitness costs without sacrificing performance, and the all-in-one genomic integration design with broad marker compatibility enhances usability. The platform’s ability to deliver large, inducible multiplexed perturbations is impactful for metabolic engineering, as shown by the 45-fold increase in succinic acid from a single transformation and precise induction. The general principles—silencing long arrays in the off state, interspersing repressor binding sites, and Csy4-based processing—should be transferable to other organisms and CRISPR systems, expanding the repertoire of orthogonal, simultaneous gene regulation programs.

This study introduces the first inducible, multiplexed CRISPRai system in S. cerevisiae that combines simultaneous activation and repression with high gRNA multiplexing (up to 24 guides) under tight, near leak-free control. The design resolves promoter-independent transcription of long arrays by array-wide silencing using a Tet-OFF repressor interleaved with gRNAs and activation via Tet-ON upon aTc addition. The authors provide an optimized, single-integration toolkit with flexible cloning and marker options, demonstrate sustained and precise inducible regulation, and achieve a 45-fold increase in succinic acid production by targeting 11 genes in central metabolism. Future work could explore alternative Cas proteins to further reduce fitness costs, extend the approach to additional chassis organisms, refine array architecture to maintain silencing at even higher guide densities, and develop activatable variants of diverse CRISPR systems to broaden applicability.

• Fitness cost: Even with tuned expression, expressing dCas9-Mxi1, dCas12a-VP, and Csy4 imposes a growth penalty relative to no-CRISPR controls. - Array design constraints: To maintain a tight off state, the system recommends no more than six gRNAs between mutTetO sites; basal leak increases with larger uninterrupted blocks. - Potential recombination: Highly repetitive arrays (particularly around Cas9 gRNA handles) can recombine over multiple passages under induction. - Induction requirements: Saturating induction used 1 µM aTc to fully release mutTetR-Mxi1. - Mechanism not fully characterized: The precise mechanisms enabling promoterless transcription of long gRNA arrays remain to be elucidated. - Generalizability pending validation: While principles should extend to other organisms and CRISPR systems, empirical validation in non-yeast hosts and with different Cas variants is needed.