Chemically-inducible CRISPR/Cas9 circuits for ultra-high dynamic range gene perturbation Authors: Rajini Srinivasan, Tao Sun, Azalia Sandles, Diana Wu, Liang Wang, Hetal Patel, Ryan Pabalate, Maram Bader, Amy Heidersbach, Clark Ho, Shiqi Xie, Andrew Ng, Benjamin Haley Published: 2025 ABSTRACT CRISPR/Cas9 technologies provide unique capabilities for modeling disease and understanding gene-to-phenotype connections. In cultured cells, chemical-mediated control of Cas9 activity can limit off-target effects and enable mechanistic study of essential genes. However, widely-used Tet-On systems often show leaky Cas9 expression, leading to unintended edits, as well as weak activity upon induction. Leakiness can be problematic in the context of Cas9 nuclease activity, which may result in cumulative DNA damage and degradation of the target cell genome over time. To overcome these deficiencies, we have established transgenic platforms that minimize Cas9 functionality in the OFF-state along with maximized and uncompromised ON-state gene editing efficiency. By combining conditional destabilization and inhibition of Cas9, we have developed an all-in-one (one or multiple guide RNAs and Cas9) ultra-tight, Tet-inducible system with exceptional dynamic range (ON vs. OFF-state) across various cell lines and targets. As an alternative to Tet-mediated induction, we have created a Branaplam-regulated splice switch module for low-baseline and robust Cas9 activity control. Lastly, for circumstances where DNA damage needs to be avoided, we have constructed a dual-control, Tet-inducible CRISPRi module for tight and potent transcriptional silencing. This upgraded suite of inducible CRISPR systems has broad applications for numerous cell types and experimental conditions. Introduction CRISPR/Cas9 is a highly adaptable and efficient genome editing tool. Commonly, users design single guide RNAs (sgRNAs) that direct Cas9 to generate double-stranded DNA (dsDNA) breaks at specified genomic target sites. Inaccurate repair of the Cas9-mediated DNA damage results in a permanent genetic modification, which allows for the disruption of individual genes or thousands in the context of functional genomic screens. The power of this technology is further amplified through methods that enable users to manipulate the timing and duration of Cas9 expression. Such control helps to minimize off-target DNA damage and facilitates the study of gene function(s) over a wide range of conditions and cellular states. Inducible gene disruption also enables mechanistic and temporal assessment of cell populations depleted of essential targets, which, due to the lethal phenotypes, present challenges with approaches that rely on constitutive Cas9 expression. However, any inducible Cas9 nuclease system is dependent on low or essentially no background editing in the off state, since this would result in an accumulation of permanent gene edits over time1,2. In addition, the post-induction on state should provide high and uniform targeting efficacy across the desired cell population. Ideally, such a system would function consistently over a wide range of cell types and target loci. Broadly applied for both in vitro and in vivo experiments, Tetracycline (Tet)/Doxycycline (Dox)-inducible systems have shown promise for timed activation of Cas9 activity. Yet, the dynamic range (e.g., contrast between the on and off state editing) of Tet-based induction is often limited by issues such as leaky background expression or sub-optimal gene targeting efficiency after induction3,4. Clonal selection from transgenic cell populations can help mitigate basal Cas9 activity, but this approach is labor-intensive, risks skewing genetic and phenotypic diversity representative of the parental cells, and may fail to fully address variability across cell types. Alternative strategies for induced gene editing have been explored, including integration of a Tet-On/Cas9 cassette with insulators separating the promoter and the reverse Tet transactivator (rtTA) elements into genomic safe harbor sites5, chemical stabilization/destabilization of degron-linked Cas94,6–9, split-Cas10, intein-Cas911, hydroxytamoxifen inducible Cas912,13 and Dox-inducible sgRNA14,15 or combinations of each4,6,8. Despite showing value, these approaches struggle with challenges related to incomplete induction and/or residual background activity, highlighting the need for technical improvement to achieve peak precision, control, and potency of genome editing. To address these limitations, we have developed a suite of enhanced, inducible CRISPR/Cas9 systems offering full on-state gene perturbation with near-absent background editing. In addition, we provide vectors that enable streamlined, single-step production of cell populations with optimal Cas9 activity profiles, eliminating the need for single-cell cloning. For general applications, we have engineered an all-in-one, ultra-tight transgene that combines Dox-inducible and degron-destabilized Cas9, a separate degron-linked anti-CRISPR module, and a constitutive sgRNA expression module to achieve ~60-95% editing efficiency and marginal (often unmeasurable) background across multiple genes and cell lines. We have further expanded the use of this system through co-expression of two sgRNAs for induced, population-level excision of a >2 kilobase (kb) genomic segment, facilitating conditional deletion of individual repetitive genomic regions, non-coding loci, and more. For circumstances where Tet/Dox-based induction must be avoided, we have created an all-in-one, chemically controlled anti-CRISPR-integrated splice switch module for Tet/Dox-independent regulation of Cas9 activity. Finally, in recognition of experimental conditions that would benefit from transcriptional silencing (e.g. CRISPR inhibition/CRISPRi) rather than directed DNA damage, we have established a Dox-inducible, degron-tagged Zim3-dCas9 expression vector that provides superior gene suppression compared to the commonly used KRAB-(KOX1) dCas9 and minimal background silencing. Together, these tools offer a marked contrast between induced and background Cas9 activity in vitro, enabling precise, on-demand CRISPR applications that bolster the flexibility and reliability of gene editing for cell-based experiments. Results The Tet-On CRISPR/Cas9 system exhibits significant basal editing To establish a baseline for background versus drug-induced editing with a foundational Tet-On Cas9 variant, we used piggyBac to stably integrate an all-in-one transgene across a panel of seven human cell lines. Here, S. pyogenes (Sp) Cas9 was placed under the control of the TRE3G Tet-inducible promoter upstream of a U6 promoter for constitutive expression of an sgRNA targeting CD81. A separate constitutive promoter (from human EF1α) was incorporated to drive expression of the reverse tetracycline transactivator (rtTA) and puromycin resistance (PuroR) gene, each separated by a T2A sequence to enable co-expression as distinct polypeptides16. Hereafter, we term this construct, inclusive of the CD81 sgRNA, pTET. Disruption of CD81, encoding a ubiquitously expressed cell surface protein, can be easily detected by a highly specific antibody and quantified through flow cytometry (Fig. 1A, Supplementary Fig. 1A). To confirm antibody specificity, we generated a CD81 knockout HEK293T line (pCAG-Cas9) with constitutive Cas9 and coincident CD81 sgRNA expression from the CAG and U6 promoters, respectively (Supplementary Fig. 1A). Wild-type and knockout cells were stained with anti-CD81 or an IgG isotype control, alongside unstained samples. Minimal signal in the isotype and unstained samples confirmed the specificity of anti-CD81 staining, which was detected in wild-type but not knockout cells. After the generation of stable polyclonal lines with pTET, cell populations were either treated with Dox to induce CD81 knockout or left untreated as a control. To ensure complete Cas9 cutting upon drug treatment and to capture cumulative leaky cutting in the untreated cells, we measured loss of CD81 via flow cytometry ten days after induction. A representative gating strategy is shown for HEK293T (Supplementary Fig. 1B). We observed significant and reproducible Dox-independent loss of CD81 in all tested cell lines (Fig. 1B). Background knockout levels ranged from a low of ~4% in K562 (bone marrow) cells to as high as 60% in SW480 (colorectal) cells. Upon treatment with Dox, robust (≥95%) depletion of CD81 was observed in all but one cell line (HCT116, ~80% depletion). Despite potent target disruption, these data highlight a pervasive problem of leaky Cas9 activity in the context of a conventional Tet-On system.Fig. 1Leaky editing with the pTET system across cell lines.A Schematic representation of the pTET piggyBac transposon plasmid (pB) and workflow depicting stable cell line generation with the piggyBac system. Cas9 expression is driven by a TRE3G Tetracycline-inducible promoter (pTRE3G). A human U6 promoter (pU6) drives constitutive expression of the CD81 sgRNA. Puromycin resistance gene (PuroR) and the reverse tetracycline transactivator (rtTA) are constitutively expressed from a human EF1α promoter and separated by a self-cleaving T2A linker. Cells that have successfully integrated the cassette are enriched by puromycin selection. Stable lines were treated with Cas9 inducing drug(s) or left untreated as a control, followed by flow cytometry to measure CD81 levels. pBO: PiggyBac transposase. B The pTET plasmid was stably integrated into HEK293T, SW480, HeLa, HAP-1, A549, HCT116 and K562 parental cells. Stable cells were treated with 2 μg/ml Doxycycline (Dox) for 10 days. Percentage knockout was determined by measuring CD81 negative cells by flow cytometry in untreated (gray bars) and Dox treated cells (colored bars); n = 3 (SW480, HeLa) or n = 4 (HCT116, HEK293T, HAP-1, A549, K562) biological replicates per stable line and treatment. Bars show mean ± SD. Source data is provided as a Source Data file. Created in BioRender. Srinivasan, R. (2025) https://BioRender.com/al5mlug. Leaky editing with the pTET system across cell lines. A Schematic representation of the pTET piggyBac transposon plasmid (pB) and workflow depicting stable cell line generation with the piggyBac system. Cas9 expression is driven by a TRE3G Tetracycline-inducible promoter (pTRE3G). A human U6 promoter (pU6) drives constitutive expression of the CD81 sgRNA. Puromycin resistance gene (PuroR) and the reverse tetracycline transactivator (rtTA) are constitutively expressed from a human EF1α promoter and separated by a self-cleaving T2A linker. Cells that have successfully integrated the cassette are enriched by puromycin selection. Stable lines were treated with Cas9 inducing drug(s) or left untreated as a control, followed by flow cytometry to measure CD81 levels. pBO: PiggyBac transposase. B The pTET plasmid was stably integrated into HEK293T, SW480, HeLa, HAP-1, A549, HCT116 and K562 parental cells. Stable cells were treated with 2 μg/ml Doxycycline (Dox) for 10 days. Percentage knockout was determined by measuring CD81 negative cells by flow cytometry in untreated (gray bars) and Dox treated cells (colored bars); n = 3 (SW480, HeLa) or n = 4 (HCT116, HEK293T, HAP-1, A549, K562) biological replicates per stable line and treatment. Bars show mean ± SD. Source data is provided as a Source Data file. Created in BioRender. Srinivasan, R. (2025) https://BioRender.com/al5mlug. An ultra-tight TET-CRISPR/Cas9 system with near-absent basal editing and high knockout efficiency In order to minimize background editing, we applied a series of modifications to pTET. These adopted different strategies to regulate Cas9 in the uninduced state including conditional stabilization of Cas9 protein, modulated anti-CRISPR expression for directed control of Cas9 activity, and drug-regulated alternative splicing to influence translation of the Cas9-containing pre-mRNA (Fig. 2A–D). The various transgenic circuits, unique control modules, and the specific drug(s) required for Cas9 expression/function are defined in Table 1. For each of these modified genetic circuits, we generated stable polyclonal cell lines in contexts that exhibited high (293T), mid (A549) and low (K562) levels of basal editing. Each stable cell population was treated with the appropriate drug or drug combination for ten days to induce CD81 depletion, while an equal number of cells were left untreated as a control. In order to obtain a performance metric for each circuit that encompasses both ON and OFF-state editing, we determined the dynamic range of Cas9 activity (i.e. the fold difference between residual editing in the OFF state and the efficiency achieved in the ON state, providing a quantitative measure of system tightness). This was calculated as the ratio of percentage CD81 knockout comparing treated versus non-treated cells.Fig. 2Development of an ultra-tight pTET system.A Schematic displaying modifications to the pTET system to minimize leaky Cas9 activity. DD: FKBP12F36VL106P destabilizing domain. DHFR: Destabilizing domain based on a mutant E.coli dihydrofolate reductase. Both fusions promote degradation of leaky Cas9 protein. AcrIIA4: anti-CRISPR gene with a nuclear localization sequence is constitutively expressed from a human PGK promoter and inhibits Cas9 activity. AcrIIA4 is linked via a GSAGSAAGSG flexible linker to a ligand-inducible degron (AcrIIA4-LID) or FKBP degron (AcrIIA4-FKBP). Xon splicing switch: cassette from SF3B3 containing the start codon in the alternatively spliced exon. B Cartoon depicting conditional stabilization of DD- or DHFR-fused Cas9. Addition of a stabilizing ligand prevents Cas9 degradation, restoring protein levels. C Cartoon illustrating the strategy to control leaky Cas9 activity with AcrIIA4. Addition of a degron-specific ligand degrades AcrIIA4, relieving Cas9 inhibition. D Splicing switch mechanism to control leaky Cas9 levels. Exon skipping prevents Cas9 translation in untreated cells. Drug-induced exon retention restores translation. E–H The pTET variants were stably transfected into HEK293T (E, F) and A549 (G, H) cells followed by drug treatment and measurement of CD81 levels as described in Fig. 1A and Table 1; (n = 3 (HEK293T; A549 pTET-Xon-Cas9) or n = 4 (A549) biological replicates per stable line and treatment). All lines express a CD81 sgRNA, except pTET-NTC (control). E, G Percentage of CD81 negative cells was determined in untreated and treated cells. Bars show mean ± SD. F, H Floating bars display the dynamic range (ratio of percentage knockout in treated vs. untreated cells) for each method; bars span minimum to maximum, the line denotes the median. Significance by two-way ANOVA (E, G), or ordinary one-way ANOVA (F, H), with Tukey’s multiple comparisons test. ns (not significant): P = 0.9988 (E); ****P < 0.0001 (E–H). I, J pTET and pTET:Ultra-tight (pTET-DD-Cas9;AcrIIA4-LID) HEK293T cells were left untreated or drug-treated for the indicated times. CD81 knockout was measured by flow cytometry (I) and targeted amplicon sequencing (J). Data represent mean ± SEM; n = 3 biological replicates per stable line, treatment and time point; n = 2 for pTET:Ultra-tight untreated sample at 240 h in (J), error bar omitted. Source data is provided as a Source Data file. Created in BioRender. Srinivasan, R. (2025) https://BioRender.com/riy5dyx.Table 1Summary of each piggyBac transgenic circuit tested in Fig. 2Transgene circuitsgRNAControl module 1Control module 2Inducing drug(s)pTET; NTCNTC––DoxpTETCD81––DoxpTET-DD-Cas9CD81DD–Dox + Shield1pTET-DHFR-Cas9CD81DHFR–Dox + TMPpTET; AcrIIA4CD81–AcrIIA4DoxpTET; AcrIIA4-FKBPCD81–AcrIIA4-FKBPDox + dTagV-1pTET; AcrIIA4-LIDCD81–AcrIIA4-LIDDox + Shield1pTET-DD-Cas9; AcrIIA4-LIDCD81DDAcrIIA4-LIDDox + Shield1pTET-Xon-Cas9CD81Xon–Dox + BranaplamFor each piggyBac plasmid, the guide RNA expressed, the control elements to minimize leaky editing, and the drugs required to induce Cas9 protein/function are listed. Unless otherwise indicated, the drugs were used at the following concentrations: 2 μg/ml Dox; 1 μM Shield1; 2 μM TMP; 1 μM dTagV-1; 20-150 nM Branaplam depending on the cell line. Development of an ultra-tight pTET system. A Schematic displaying modifications to the pTET system to minimize leaky Cas9 activity. DD: FKBP12F36VL106P destabilizing domain. DHFR: Destabilizing domain based on a mutant E.coli dihydrofolate reductase. Both fusions promote degradation of leaky Cas9 protein. AcrIIA4: anti-CRISPR gene with a nuclear localization sequence is constitutively expressed from a human PGK promoter and inhibits Cas9 activity. AcrIIA4 is linked via a GSAGSAAGSG flexible linker to a ligand-inducible degron (AcrIIA4-LID) or FKBP degron (AcrIIA4-FKBP). Xon splicing switch: cassette from SF3B3 containing the start codon in the alternatively spliced exon. B Cartoon depicting conditional stabilization of DD- or DHFR-fused Cas9. Addition of a stabilizing ligand prevents Cas9 degradation, restoring protein levels. C Cartoon illustrating the strategy to control leaky Cas9 activity with AcrIIA4. Addition of a degron-specific ligand degrades AcrIIA4, relieving Cas9 inhibition. D Splicing switch mechanism to control leaky Cas9 levels. Exon skipping prevents Cas9 translation in untreated cells. Drug-induced exon retention restores translation. E–H The pTET variants were stably transfected into HEK293T (E, F) and A549 (G, H) cells followed by drug treatment and measurement of CD81 levels as described in Fig. 1A and Table 1; (n = 3 (HEK293T; A549 pTET-Xon-Cas9) or n = 4 (A549) biological replicates per stable line and treatment). All lines express a CD81 sgRNA, except pTET-NTC (control). E, G Percentage of CD81 negative cells was determined in untreated and treated cells. Bars show mean ± SD. F, H Floating bars display the dynamic range (ratio of percentage knockout in treated vs. untreated cells) for each method; bars span minimum to maximum, the line denotes the median. Significance by two-way ANOVA (E, G), or ordinary one-way ANOVA (F, H), with Tukey’s multiple comparisons test. ns (not significant): P = 0.9988 (E); ****P < 0.0001 (E–H). I, J pTET and pTET:Ultra-tight (pTET-DD-Cas9;AcrIIA4-LID) HEK293T cells were left untreated or drug-treated for the indicated times. CD81 knockout was measured by flow cytometry (I) and targeted amplicon sequencing (J). Data represent mean ± SEM; n = 3 biological replicates per stable line, treatment and time point; n = 2 for pTET:Ultra-tight untreated sample at 240 h in (J), error bar omitted. Source data is provided as a Source Data file. Created in BioRender. Srinivasan, R. (2025) https://BioRender.com/riy5dyx. Table 1Summary of each piggyBac transgenic circuit tested in Fig. 2Transgene circuitsgRNAControl module 1Control module 2Inducing drug(s)pTET; NTCNTC––DoxpTETCD81––DoxpTET-DD-Cas9CD81DD–Dox + Shield1pTET-DHFR-Cas9CD81DHFR–Dox + TMPpTET; AcrIIA4CD81–AcrIIA4DoxpTET; AcrIIA4-FKBPCD81–AcrIIA4-FKBPDox + dTagV-1pTET; AcrIIA4-LIDCD81–AcrIIA4-LIDDox + Shield1pTET-DD-Cas9; AcrIIA4-LIDCD81DDAcrIIA4-LIDDox + Shield1pTET-Xon-Cas9CD81Xon–Dox + BranaplamFor each piggyBac plasmid, the guide RNA expressed, the control elements to minimize leaky editing, and the drugs required to induce Cas9 protein/function are listed. Unless otherwise indicated, the drugs were used at the following concentrations: 2 μg/ml Dox; 1 μM Shield1; 2 μM TMP; 1 μM dTagV-1; 20-150 nM Branaplam depending on the cell line. Summary of each piggyBac transgenic circuit tested in Fig. 2 For each piggyBac plasmid, the guide RNA expressed, the control elements to minimize leaky editing, and the drugs required to induce Cas9 protein/function are listed. Unless otherwise indicated, the drugs were used at the following concentrations: 2 μg/ml Dox; 1 μM Shield1; 2 μM TMP; 1 μM dTagV-1; 20-150 nM Branaplam depending on the cell line. Unless otherwise noted, inducer concentrations (2 µg/ml doxycycline, 1 µM Shield1, 1 µM dTAGV-1, 2 µM trimethoprim, or 20–150 nM Branaplam depending on cell type) were pre-titrated in reporter assays to maximize performance while maintaining viability, and these doses align with ranges previously reported (see Methods). Broadly, the re-engineered systems provided a substantial decrease in background editing and, consequently, a general improvement in dynamic range (Fig. 2E–H). Across tested lines, expression of a non-targeting control sgRNA from the pTET-Cas9 vector (pTET-NTC) demonstrated that CD81 levels were unaffected solely by Dox addition (Fig. 2E, G, Supplementary Fig. 2A). For several conditions, basal CD81 disruption in cell lines with high (HEK293T, Fig. 2E) and moderate (A549, Fig. 2G) leakiness was reduced to levels similar to that seen with pTET-NTC. K562 cells, which already demonstrated a low background similar to pTET-NTC, and high Dox-induced editing with pTET, did not benefit from additional vector modification (Supplementary Fig. 2A, B). While there was a general improvement in the dynamic range of editing across the tested conditions, relative to pTET, the peak editing efficiency upon induction was variable; detailed below. Similar to previous studies, we genetically fused conditional destabilizing degron domains17,18 to Cas9 within the pTET vector context, which would lead to rapid turnover of leaky Cas9 protein in the absence of Dox and the small molecule used to stabilize each degron (Fig. 2B, Table 1)6. Here, DD references a double mutant variant of the FKBP12 protein (FKBP12 F36VL106P) that is continuously degraded until treated with Shield117. DHFR is an E.coli dihydrofolate reductase mutant that destabilizes tagged proteins in the absence of Trimethoprim (TMP)18. Although the destabilizing domain-tagged Cas9 variants (pTET-DD-Cas9 and pTET-DHFR-Cas9) displayed comparable percentage knockout in the ON state, background editing was lower with pTET-DD-Cas9 versus pTET-DHFR-Cas9. This resulted in higher dynamic range of basal versus induced activity with the DD-tagged Cas9 module (Fig. 2F, H). The discovery of anti-CRISPR proteins, several of which can function as natural inhibitors of Cas9 activity, has inspired the development of systems for titratable Cas9 deactivation2,19–21. To investigate the potential application of anti-CRISPRs for reducing basal editing, we modified the pTET vector to constitutively express engineered variants of AcrIIA4, previously identified as a highly effective inhibitor of SpCas919,22. For this, we assessed AcrIIA4 fused to a Nucleoplasmin NLS and, as an additional layer of control, tested C-terminal fusions of FKBP12F36V or ligand-inducible (LID) conditional degrons (Fig. 2C). FKBP12F36V (hereafter referred to as FKBP) can be used for selective degradation upon treatment with an appropriate dTag ligand23,24. LID is an engineered variant of FKBP with a cryptic degron at the C-terminus, and is degraded in response to treatment with Shield125. Co-treatment with Dox and dTag (i.e. dTagV-1) or Shield1 is expected to induce Cas9 expression and in parallel degrade the AcrIIA4, significantly shifting the stoichiometric balance in favor of Cas9 protein/function (Fig. 2C, Table 1). Accordingly, in the OFF state, continuous expression of non-degradable AcrIIA4 (pTET;AcrIIA4) virtually abolished basal editing (Fig. 2E, G). However, Dox-induced targeting was compromised with this variant in some cellular contexts (see A549 cells, Fig. 2G). Presumably, the induced levels of Cas9 were insufficient to overcome the suppressive effects of the constitutively expressed anti-CRISPR protein in these contexts. Consistent with this notion, editing efficiency was largely restored and/or consistently high in pTET;AcrIIA4-FKBP and pTET;AcrIIA4-LID, the degron controlled AcrIIA4 lines. In HEK293T and A549 cells, there was no significant difference in drug-induced CD81 disruption between the two degron-tagged AcrIIA4 systems (Fig. 2E, G). In K562 cells, despite a ~ 10% higher editing efficiency with pTET;AcrIIA4-FKBP compared to pTET;AcrIIA4-LID, the dynamic range was similar due to marginally greater background gene disruption with the AcrIIA4-FKBP fusion (Supplementary Fig. 2A, B). While subtle differences were observed across cellular contexts for pTET-AcrIIA4-FKBP and pTET-AcrIIA4-LID, the directionality was inconsistent, and we determined these systems to be functionally interchangeable with one another and the pTET-DD-Cas9 variant. Targeted degradation and direct inhibition of Cas9 protein both proved to be effective for reducing basal gene editing; we next evaluated whether we could achieve similar results through conditional translation of Cas9. For this, we exploited the Xon alternative splicing switch26 based on a modified, Branaplam-sensitive splicing cassette from SF3B3. Placed within the pTET vector (pTET-Xon-Cas9), immediately downstream of the TRE3G promoter, the modified cassette positions a small exon with an initiator (AUG) codon within an SF3B3 intron segment. In the absence of Branaplam, the exon with the initiator codon will be spliced over (i.e. excluded) and Cas9 translation will be impaired (Fig. 2D). Upon co-treatment with Dox and Branaplam, a switch in the splicing mechanism will retain the AUG-containing exon, allowing in-frame and full-length production of Cas9 protein. Across all tested cell lines, basal editing was reduced with pTET-Xon-Cas9 relative to pTET, but this came at a severe cost to perturbation once activated (Fig. 2E, G, Supplementary Fig. 2A). These results suggested that additional measures were needed to optimize the performance and compatibility of the Xon and pTET systems. Encouraged by the targeting efficacy after induction with either the degradable Cas9 or anti-CRISPR control modules, and the potential for further reduction in basal editing, we generated a composite pTET vector with both the DD-Cas9 and AcrIIA4-LID systems (pTET-DD-Cas9;AcrIIA4-LID). Using this design, co-treatment with Dox and Shield1 will induce and stabilize Cas9, while simultaneously relieving inhibition of Cas9 activity through depletion of AcrIIA4. We favored this specific combination of control modules, since it restricts the number of inducing agents to two rather than the three required for incorporation of either DHFR-Cas9 or AcrIIA4-FKBP. Application of pTET-DD-Cas9;AcrIIA4-LID resulted in a > 100-fold dynamic range when comparing the ON versus OFF states for CD81 depletion in both the HEK293T and A549 backgrounds, which was 2-fold greater than any other system in these contexts, with a trend towards improved dynamic range in K562 cells (Fig. 2F, H, Supplementary Fig. 2A, B). This was due to the low background editing level, which was indistinguishable from the pTET-NTC control. Induced editing remained high (>80% within polyclonal cell populations) with pTET-DD-Cas9;AcrIIA4-LID, but with a marginal (~10%) decrease relative to pTET in both the A549 and K562 cells. We therefore decided to prioritize and expand our evaluation of the pTET-DD-Cas9;AcrIIA4-LID system (hereafter named pTET:Ultra-tight) due to its unique capacity for nearly eliminating background activity with minimal impact on induced editing. To further validate the function of the control modules, we performed western immunoblotting to assess Cas9 protein expression in pTET and pTET:Ultra-tight systems following drug treatments (Supplementary Fig. 3A). HEK293T cells transfected with each construct were treated with the indicated small molecules for 72 h, followed by protein extraction and immunoblotting for Cas9. As expected, Cas9 protein was undetectable in untreated pTET:Ultra-tight cells, reflecting tight regulation. Interestingly, Cas9 protein levels remained below the detection limit in pTET cells despite the significant background editing observed by flow cytometry, suggesting that functional activity can occur at levels undetectable by western blot analysis. As such, target gene depletion is a more sensitive readout of leaky Cas9 activity relative to Cas9 protein detection. Cas9 protein was readily observed upon Dox treatment of both pTET and pTET:Ultra-tight cells. Expression was lower in the pTET:Ultra-tight condition, likely due to degradation via the DD tag in the absence of Shield1. Cas9 protein was not seen when pTET:Ultra-tight cells were treated with Shield1 alone. Co-treatment of pTET:Ultra-tight cells with Dox and Shield1 increased Cas9 protein levels, though to a lesser extent than in pTET cells. However, this expression level was sufficient to drive efficient target gene knockout as indicated by flow cytometry results (Fig. 2E–H, Supplementary Fig. 2A, B). Attempts to detect Anti-CRISPR protein by immunoblotting were unsuccessful due to limitations of available antibodies. Despite this, changes in CD81 protein levels were consistent with the expected regulation of the degron-tagged Anti-CRISPR in the presence or absence of the specific drugs (Fig. 2E–H). Time-resolved target depletion demonstrates concordance between protein loss and indel formation Having established pTET:Ultra-tight as a more tightly regulated system than pTET, we next compared the kinetics of editing for each system over a period of 10 days (Fig. 2I). Untreated pTET cells exhibited ~50% CD81 depletion, which did not increase further over time. In contrast, untreated pTET:Ultra-tight displayed minimal basal editing throughout the time course. In drug treated pTET cells, maximal CD81 depletion occurred by 72 h. pTET:Ultra-tight cells took longer than pTET cells to achieve maximal CD81 depletion, but reached equivalent levels by 144-240 h. To confirm that flow cytometry results reflected underlying genomic changes, we performed targeted amplicon sequencing of CD81 on the same samples to quantify indel frequencies over time (Fig. 2J). The amplicon sequencing data closely paralleled target protein levels. Untreated pTET cells displayed ~50% indel generation at the CD81 locus throughout the time course. In pTET:Ultra-tight cells, indels remained minimal without treatment, confirming suppression of leaky Cas9 activity. For both systems, between 6 and 48 h after treatment, indel frequencies exceeded the percentage of CD81-depleted cells, suggesting a lag between genomic editing and observable protein loss, likely due to transcriptional/translational rates and/or protein half-life. Similar to flow cytometry results, upon induction, pTET:Ultra-tight cells showed a modest delay compared to pTET cells before achieving maximal editing. Indel frequencies plateaued in both systems by 144 h of drug exposure. A CD81 knockout control line (Supplementary Fig. 1A) was established by stable genomic integration of a constitutive Cas9 and CD81 sgRNA-expressing transgene. This line exhibited complete protein depletion and genomic indel formation by 24 h after removal of puromycin (used for selecting stable integrants) and these levels remained constant through the duration of our assay (~240 h). (Supplementary Fig. 3B, C). Together, these time-course experiments confirmed the minimal background editing and high efficiency with the pTET:Ultra-tight system as quantified by both indel generation and protein depletion. While our initial assays focused on a 5–10-day window post-induction, the ability to detect robust editing at earlier time points expands the utility of the system, allowing researchers to capture transient phenotypes, resolve early molecular events, or study genes whose prolonged loss impairs cell viability. Consistent pTET:Ultra-tight functionality across a diverse cell line panel In order to assess the broad applicability of pTET:Ultra-tight across an expanded set of cell lines, we stably integrated pTET and pTET:Ultra-tight into SW480, HeLa, HAP-1, and HCT116. In our initial evaluation of pTET, these four cell lines exhibited a range of basal editing from ~60% to 8% (Fig. 1B). Consistent with our previous observations, the pTET:Ultra-tight lines displayed substantially reduced basal editing in untreated cells relative to pTET in all cell types (Fig. 3A). Of the tested lines, SW480 cells represented an outlying context, where pTET:Ultra-tight was unable to fully reduce background editing. We therefore decided to compare pTET:Ultra-tight to an expanded set of genetic circuits (Supplementary Fig. 4A), revealing that it remained the superior option for balancing ON versus OFF-state editing (Supplementary Fig. 4B). Upon drug exposure, the pTET:Ultra-tight system maintained a knockout efficiency greater than 90% in all but one cancer cell line; HCT116 (58%). HCT116 cells also displayed the lowest induced knockout rate (~80%) tested across all lines with the pTET circuit, suggesting the Dox-regulated circuit requires further optimization within this particular cellular context (Fig. 1B and Fig. 3A). Despite a reduction in knockout efficiency within HCT116 cells with the pTET:Ultra-tight system, the gains in dynamic range between the ON and OFF states, due to a near-complete elimination of background editing, were substantial (>50-fold relative to pTET); a trend that extended to the additional lines tested in this study (Fig. 3B).Fig. 3Ultra-tight pTET exhibits stringent regulation across diverse cell types.A, B pTET and pTET:Ultra-tight constructs were stably integrated into the indicated cell lines. Stable cells were left untreated (gray bars) or treated (colored bars) for 10 days with 2 μg/ml Dox (pTET) or 2 μg/ml Dox + 1 μM Shield1 (pTET:Ultra-tight) to induce Cas9 expression, followed by measurement of CD81 levels as described in Fig. 1A (n = 3 biological replicates per stable line and treatment). A Bars show mean ± SD. B Floating bars display the ratio of percentage CD81 knockout in treated vs. untreated cells with pTET and pTET:Ultra-tight systems for each cell line; bars extend from the minimum to maximum values, and the line denotes the median. Significance by two-way ANOVA with Tukey’s multiple comparisons test (A), and unpaired two-tailed Welch’s t-test (B). ns: P > 0.9999 (A; SW480), P = 0.0798 (A; HAP-1); ****P < 0.0001 (A); **P < 0.01 (B; SW480, P = 0.0010; HAP-1, P = 0.0047; HeLa, P = 0.0035; HCT116, P = 0.0030; KOLF2.1J, P = 0.0046). Source data is provided as a Source Data file. Created in BioRender. Srinivasan, R. (2025) https://BioRender.com/hujprvj. Ultra-tight pTET exhibits stringent regulation across diverse cell types. A, B pTET and pTET:Ultra-tight constructs were stably integrated into the indicated cell lines. Stable cells were left untreated (gray bars) or treated (colored bars) for 10 days with 2 μg/ml Dox (pTET) or 2 μg/ml Dox + 1 μM Shield1 (pTET:Ultra-tight) to induce Cas9 expression, followed by measurement of CD81 levels as described in Fig. 1A (n = 3 biological replicates per stable line and treatment). A Bars show mean ± SD. B Floating bars display the ratio of percentage CD81 knockout in treated vs. untreated cells with pTET and pTET:Ultra-tight systems for each cell line; bars extend from the minimum to maximum values, and the line denotes the median. Significance by two-way ANOVA with Tukey’s multiple comparisons test (A), and unpaired two-tailed Welch’s t-test (B). ns: P > 0.9999 (A; SW480), P = 0.0798 (A; HAP-1); ****P < 0.0001 (A); **P < 0.01 (B; SW480, P = 0.0010; HAP-1, P = 0.0047; HeLa, P = 0.0035; HCT116, P = 0.0030; KOLF2.1J, P = 0.0046). Source data is provided as a Source Data file. Created in BioRender. Srinivasan, R. (2025) https://BioRender.com/hujprvj. To confirm that protein-level changes were reflected at the genomic locus, we evaluated indel generation in HAP-1 cells by targeted amplicon sequencing. We observed strong agreement between protein-level and genomic changes (Supplementary Fig. 4C), mirroring the concordance seen in our kinetic analysis. In addition to immortalized cell lines, we extended our evaluation to include a more sensitive, non-transformed line, KOLF2.1J, an induced pluripotent stem cell (iPSC) line that has been extensively characterized and favored for its genomic stability, efficient CRISPR/Cas9 editing, and robust differentiation capacity27,28. Stable, polyclonal populations were engineered with pTET or pTET:Ultra-tight and treated with either Dox (pTET) or Dox + Shield1 (pTET:Ultra-tight) for 10 days followed by flow cytometry to measure CD81 levels. Both treated and untreated KOLF2.1J stable cell lines exhibited morphological features characteristic of healthy iPSCs, including prominent nuclei, tightly packed colony growth with well-defined borders, and minimal evidence of spontaneous differentiation (Supplementary Fig.4D). Additionally, SSEA-4, a well-established marker of iPSC pluripotency, was highly expressed across all conditions, confirming that the cells maintained a pluripotent state irrespective of transgene or drug treatment (Supplementary Fig. 4E). While we observed a modest decrease in targeting efficiency when comparing pTET (~93%) to pTET:Ultra-tight (~78%), background editing was absent with pTET:Ultra-tight, resulting in the highest dynamic range of editing for all tested cell types (>200-fold difference between the ON vs. OFF-state). Notably, the ~78% editing efficiency observed with pTET:Ultra-tight in KOLF2.1J represents a substantial improvement over published inducible systems in stem cells, which report efficiencies closer to ~40%5,6. Together, these results suggest pTET:Ultra-tight consistently outperforms pTET across multiple cellular contexts. The pTET:Ultra-tight system maintains highly controlled regulation at additional target genes To investigate whether the pTET:Ultra-tight system maintains stringent regulation at target loci other than CD81, we replaced the CD81 sgRNA in pTET and pTET:Ultra-tight constructs with sgRNAs targeting an expanded cohort of non-essential cell surface markers; β2-microglobulin (B2M), CD298, and CD9. As previously, we generated stable, polyclonal HEK293T cell lines with these constructs, treated each line with either Dox or Dox + Shield1, and evaluated target protein levels 10 days later by antibody staining and flow cytometry. Untreated cells were included as a control to determine basal editing. Consistent with our observations for CD81, we detected background editing in untreated cells at all three target loci with the pTET system (Fig. 4A–C). In contrast, pTET:Ultra-tight displayed strikingly diminished basal editing for each target. Importantly, upon drug exposure, population-wide knockout efficiencies >90% were observed for all target genes. Significant improvement in dynamic range was observed across targets with the pTET:Ultra-tight system (Fig. 4D–F). Consistent with prior observations, editing at the B2M locus showed close alignment between indel frequencies and protein loss (Supplementary Fig. 5).Fig. 4Ultra-tight system outperforms pTET at additional target loci.pTET or pTET:Ultra-tight constructs expressing sgRNA targeting B2M (A, D), CD298 (B, E) or CD9 (C, F) were stably integrated into HEK293T cells. Stable cells were left untreated or treated for 10 days with 2 μg/ml Dox (pTET) or 2 μg/ml Dox + 1 μM Shield1 (pTET:Ultra-tight) to induce Cas9 expression (n = 3 biological replicates per stable line and treatment for CD298 and CD9, and n = 6 biological replicates for B2M). Percentage of CD81 negative cells was measured as described in Fig. 1A. A–C Bars show mean ± SD. D–F Floating bars display ratio of percentage knockout comparing treated vs. untreated cells for pTET and pTET:Ultra-tight at each target locus; bars extend from the minimum to maximum values, and the line denotes the median. Significance by two-way ANOVA adjusted with Tukey’s multiple comparisons test (A–C) or unpaired two-tailed Welch’s t-test (D–F). ns: P = 0.3259 (A), P = 0.4739 (B); *P < 0.05 (C, P = 0.0448); **P < 0.01 (E, P = 0.0090; F, P = 0.0027); ***P < 0.001 (D, P = 0.0001); ****P < 0.0001 (A–C). Source data is provided as a Source Data file. Created in BioRender. Srinivasan, R. (2025) https://BioRender.com/coigpze. Ultra-tight system outperforms pTET at additional target loci. pTET or pTET:Ultra-tight constructs expressing sgRNA targeting B2M (A, D), CD298 (B, E) or CD9 (C, F) were stably integrated into HEK293T cells. Stable cells were left untreated or treated for 10 days with 2 μg/ml Dox (pTET) or 2 μg/ml Dox + 1 μM Shield1 (pTET:Ultra-tight) to induce Cas9 expression (n = 3 biological replicates per stable line and treatment for CD298 and CD9, and n = 6 biological replicates for B2M). Percentage of CD81 negative cells was measured as described in Fig. 1A. A–C Bars show mean ± SD. D–F Floating bars display ratio of percentage knockout comparing treated vs. untreated cells for pTET and pTET:Ultra-tight at each target locus; bars extend from the minimum to maximum values, and the line denotes the median. Significance by two-way ANOVA adjusted with Tukey’s multiple comparisons test (A–C) or unpaired two-tailed Welch’s t-test (D–F). ns: P = 0.3259 (A), P = 0.4739 (B); *P < 0.05 (C, P = 0.0448); **P < 0.01 (E, P = 0.0090; F, P = 0.0027); ***P < 0.001 (D, P = 0.0001); ****P < 0.0001 (A–C). Source data is provided as a Source Data file. Created in BioRender. Srinivasan, R. (2025) https://BioRender.com/coigpze. In general, the robust agreement observed between targeted amplicon sequencing and flow cytometry across cell types, target loci, and experimental contexts supports pTET:Ultra-tight as a superior system for regulating Cas9 activity. pTET:Ultra-tight system exhibits tight control over essential gene knockout Large-scale functional genomic screening campaigns have revealed a host of core and condition-specific essential genes29. Detailed mechanistic study of these factors is advantaged with an experimental system that allows precise control over the timing of gene disruption, as constitutive knockout is expected to result in early growth arrest or cell loss. Given the near-absent leakiness and potent induction observed with the pTET:Ultra-tight system, we sought to evaluate its potential for investigating an essential gene phenotype. We constructed pTET and pTET:Ultra-tight all-in-one plasmids expressing an sgRNA targeting the Polo-like kinase 1 (PLK1) gene. PLK1 is a serine/threonine kinase and a master regulator of cell cycle and mitosis. Disruption of PLK1 results in proliferation defects. PLK1 is overexpressed in several cancers and is considered a potential target for cancer therapy30,31. Each of the PLK1-targeting constructs, as well as a pTET vector expressing a NTC sgRNA, were stably integrated into HEK293T cells to create polyclonal populations. HEK293T cells were chosen based on the relatively high rate of background editing with pTET (Fig. 1B). Each cell population was either treated with Dox (pTET) or Dox + Shield1 (pTET:Ultra-tight) to induce Cas9 activity or left untreated as a control for background (leaky) targeting. Cell growth was monitored by live imaging over five days in the presence/absence of drugs (Fig. 5A). Stable lines were successfully generated for all constructs, but there were notable differences in growth rates in the presence and absence of drug treatment. In the untreated state, the pTET line targeting PLK1 exhibited reduced proliferation, suggesting knockout of the target gene in the absence of induction (Fig. 5B). Conversely, the growth rate of the pTET:Ultra-tight line expressing PLK1 sgRNA was comparable to the pTET-NTC control line (Fig. 5B). Upon drug treatment, growth was almost fully inhibited in both pTET and pTET:Ultra-tight lines, although the latter demonstrated a greater reduction in growth rate. As previously shown with non-essential gene disruption (Fig. 2E–H, Fig. 4A–F), the pTET:Ultra-tight system demonstrated a significantly improved dynamic range between treated and untreated states, as compared to pTET (Fig. 5C) While a population-level phenotype was observed with both pTET and pTET:Ultra-tight, we interpret the difference in dynamic range between the systems to reflect “pruning” of cells with leaky and, likely, the greatest capacity for PLK1 disruption in the pTET context. In that scenario, cells with diminished Cas9 activity would survive but induction would result in a less penetrant phenotype. Our data reinforce the notion that the pTET:Ultra-tight system supersedes pTET for implementing effective, population-level editing control across multiple cell lines and targets including essential genes.Fig. 5pTET:Ultra-tight system exhibits tight control over essential gene knockout and deletion of a ~ 2.5 kb genomic region.A Schematic overview of the cell growth assay workflow. Stable lines were generated in HEK293T with pTET and pTET:Ultra-tight constructs expressing a PLK1 guide RNA, while pTET-NTC expresses a control guide RNA. An equal number of cells were plated with or without drug treatment and imaged in an Incucyte over 5 days in 4 h intervals (n = 3 biological replicates per stable line, treatment and time point). B Growth curves are shown for each line in the presence and absence of drug treatment. Cells were treated with 2 μg/ml Dox (pTET and pTET-NTC), 2 μg/ml Dox + 1 μM Shield1 (pTET:Ultra-tight), or left untreated. Data represent mean values ± SEM. C Floating bars display change in percentage confluence for each line at 120 h; bars extend from the minimum to maximum values, and the line denotes the median. Significance by unpaired two-tailed Welch’s t-test. **p < 0.01 (P = 0.0099). D Top: Schematic of the human H3-3A locus showing the sgRNA cut sites and expected amplicon products from wild-type and deleted alleles. Filled blue squares represent exons, with exons 2 and 3 depicted as dark blue squares. Half arrows represent PCR primers. Bottom: PCR gel image of amplicon products from the indicated cell lines are shown (n = 1 biological replicate per condition). Cells were either left untreated or treated with 2 μg/ml Dox (parental, pTET, pTET-NTC) or 2 μg/ml Dox + 1 μM Shield1 (pTET:Ultra-tight) for 72 h. Source data is provided as a Source Data file. Created in BioRender. Srinivasan, R. (2025) https://BioRender.com/o0erd6s. pTET:Ultra-tight system exhibits tight control over essential gene knockout and deletion of a ~ 2.5 kb genomic region. A Schematic overview of the cell growth assay workflow. Stable lines were generated in HEK293T with pTET and pTET:Ultra-tight constructs expressing a PLK1 guide RNA, while pTET-NTC expresses a control guide RNA. An equal number of cells were plated with or without drug treatment and imaged in an Incucyte over 5 days in 4 h intervals (n = 3 biological replicates per stable line, treatment and time point). B Growth curves are shown for each line in the presence and absence of drug treatment. Cells were treated with 2 μg/ml Dox (pTET and pTET-NTC), 2 μg/ml Dox + 1 μM Shield1 (pTET:Ultra-tight), or left untreated. Data represent mean values ± SEM. C Floating bars display change in percentage confluence for each line at 120 h; bars extend from the minimum to maximum values, and the line denotes the median. Significance by unpaired two-tailed Welch’s t-test. **p < 0.01 (P = 0.0099). D Top: Schematic of the human H3-3A locus showing the sgRNA cut sites and expected amplicon products from wild-type and deleted alleles. Filled blue squares represent exons, with exons 2 and 3 depicted as dark blue squares. Half arrows represent PCR primers. Bottom: PCR gel image of amplicon products from the indicated cell lines are shown (n = 1 biological replicate per condition). Cells were either left untreated or treated with 2 μg/ml Dox (parental, pTET, pTET-NTC) or 2 μg/ml Dox + 1 μM Shield1 (pTET:Ultra-tight) for 72 h. Source data is provided as a Source Data file. Created in BioRender. Srinivasan, R. (2025) https://BioRender.com/o0erd6s. Conditional deletion of a ~ 2.5 kb genomic locus with paired guide RNAs Thus far, we have explored the general capability of pTET:Ultra-tight for conditional disruption of genes using individual, coding exon-targeting sgRNAs. The use of paired sgRNAs, where two guides are co-expressed and target positions spaced along the same chromosome, enables specified excision of large regions of DNA, enabling more complex and expanded genomic studies beyond what is capable with single sgRNA targeting. Such edits could include deletion of regulatory DNA elements (i.e. enhancers), non-coding RNAs, repetitive DNA, specific exons, and genes with family members or pseudogenes sharing highly similar exonic sequences32–36. To this end, we assessed the potential for pTET:Ultra-tight and paired sgRNA expression to induce targeted genomic deletions. As proof-of-concept for conditional genomic excision with pTET:Ultra-tight, we focused on the H3-3A locus, which expresses the non-canonical H3.3 A histone variant. H3-3A and the near-identical H3-3B genes present a distinct use case for targeted deletion, since these share 100% nucleotide similarity across their coding sequences, while the intron and untranslated regions are unique37,38. In addition, H3-3A has several pseudogenes with >90% sequence overlap with the H3-3A open reading frame39. These uncommon features preclude the use of single sgRNAs for the disruption of H3-3A via exonic targeting. However, taking advantage of the non-duplicated genomic segments, we designed a pair of sgRNAs targeting intronic sequences flanking exons 2 and 3, which span ~2.5 kilobases (kb), of the H3-3A locus (Fig. 5D). For expression of the H3-3A-targeting sgRNA pair, we cloned one guide within the original U6 promoter cassette in our pTET and pTET:Ultra-tight vectors, while the second guide of the pair was cloned downstream of a human H1 promoter. Stable, polyclonal lines were generated in HEK293T with both the pTET and pTET:Ultra-tight constructs, and cells were either left untreated or exposed to inducing drug(s) for 72 h. After this period, we collected genomic DNA and performed PCR with primers spanning the target sites. An amplicon size of ~2.9 kb indicated the presence of the wild-type allele, while successful deletion of the target locus would result in a shorter amplicon of ~0.4 kb. In the absence of drug treatment, the ~2.9 kb product corresponding to the wild-type allele was detected in parental, pTET-NTC control, pTET and pTET:Ultra-tight cells (Fig. 5D). However, only within the pTET cells did we observe the ~0.4 kb deletion product, indicating leaky editing in this context. Upon drug treatment, the deletion product was observed for both pTET and pTET:Ultra-tight cell lines with a concomitant reduction in signal for the full-length amplicon. Taken together, these results demonstrate the technical feasibility of this approach and validate the use of pTET:Ultra-tight and paired sgRNA expression for precisely controlled deletion of genomic loci. Development of an enhanced, Dox-independent, conditional CRISPR/Cas9 system While the Tet/Dox system is commonly employed for conditional transgene expression, it is not uniformly compatible with every cell model or experimental situation. For instance, Dox can disrupt mitochondrial function, making the Tet/Dox system less desirable for studies involving this organelle40,41. Separately, orthogonal control of distinct genetic circuits, to direct or understand complex cellular processes, may be enabled through the use of non-Tet/Dox systems for modulating CRISPR/Cas9 activity42. Furthermore, the ability to orthogonally control two processes in the same cell model is a common need that necessitates exploration of an alternative inducible system for controlling CRISPR activity43,44. Previous studies have applied constitutive Cas9 expression in combination with conditional degradation or alternative splicing for drug-regulated gene editing8,45,46. To evaluate the potency and background editing of Dox-independent induction of Cas9, we designed all-in-one piggyBac vectors with Cas9 expression driven by a constitutive promoter (CMV) and protein levels controlled by either DD or DHFR fusion domains or a Xon cassette (Supplementary Fig. 6A). In addition, these vectors contained a constitutive CD81-targeting sgRNA expression cassette. Stable, polyclonal HEK293T cell lines were generated for each construct. After antibiotic selection, cell populations were cultured for 10 days with or without respective drug treatments. DD- (N or C-terminal) and DHFR-fused Cas9 expression resulted in 100% depletion of CD81 in the presence or absence of drug treatment, indicating a complete lack of Cas9 activity control under these conditions (Supplementary Fig. 6B). By contrast, the Xon-Cas9 construct enabled inducible control of Cas9 activity, with ~88% disruption of CD81 in the presence of Branaplam. However, this system displayed nearly 20% basal editing in the untreated HEK293T cell context, which encouraged further optimization (Supplementary Fig. 6B). Based on our ability to mitigate leaky Cas9 activity within the pTET system, we deployed similar strategies to improve upon the Xon-Cas9 (Fig. 6A). To begin, we added the DD or DHFR domain to the N- or C-terminus of Cas9 under control of Xon. Our data with the optimized TET systems suggested that N-terminal fusion of DD and DHFR would reduce background editing activity (Fig. 2E, G). However, in the context of the Xon switch, an N-terminal degron fusion led to 100% leakiness of Cas9 activity (Supplementary Fig. 6C). We reasoned that this could be due to an internal start codon provided by the DD or DHFR domains, which would enable Cas9 translation without Branaplam-induced exon inclusion (Supplementary Fig. 6D). Consistent with this hypothesis, C-terminal DD or DHFR domain fusions demonstrated a substantial reduction in background effects without impacting ON state editing (Fig. 6B). We next tested the addition of an anti-CRISPR protein AcrIIA4 with or without the FKBP degron. The addition of AcrIIA4 dramatically reduced the leakiness of Xon-Cas9 in HEK293T cells, but at the cost of significantly compromised inducible cutting (Fig. 6B). By contrast, inclusion of the AcrIIA4-FKBP control module not only reduced the leakiness but also maintained full ON state editing, offering the best dynamic range of all conditions tested in this cell type (Fig. 6B, C).Fig. 6Development of an enhanced Dox-independent inducible Cas9.A Schematic diagram of the Xon based ultra-tight inducible Cas9. Destabilizing domain DD or DHFR was fused to the C-terminus of Cas9 to reduce the basal expression from the splicing switch. Alternatively, anti-CRISPR proteins with or without the FKBP degron were expressed from a human PGK promoter to inhibit Cas9 activity. The various Xon-Cas9 piggyBac constructs were stably transfected into HEK293T cells followed by drug treatment and measurement of CD81 levels as described in Fig. 1A and Table 1 (n = 3 biological replicates per stable line and treatment). B CD81 knockout efficiency was determined in the untreated (gray bars) and treated (colored bars) HEK293T cells. Bars show mean ± SD. C Floating bars show the dynamic range as the ratio of percentage CD81 disruption in treated vs. untreated cells for each method; bars extend from the minimum to maximum values, and the line denotes the median. D Time-course analysis of CD81 knockout in Xon-Cas9 vs. Xon-Cas9;AcrIIA4-FKBP HEK293T cells, that were left untreated or treated with either 20 nM Branaplam (Xon-Cas9) or 20 nM Branaplam + 1 μM dTagV-1 (Xon-Cas9;AcrIIA4-FKBP). Percentage CD81 negative cells were quantified by flow cytometry at the indicated time points (n = 3 biological replicates per stable line, treatment and time point). Data represent mean ± SEM. E, F K562 cells were stably integrated with the different versions of Xon based inducible Cas9 (n = 3 biological replicates per stable line and treatment). Bar graphs display CD81 perturbation efficiency in untreated (gray bars) and treated (colored bars, 150 nM Branaplam + 1 μM dTagV-1) cells. Bars show mean ± SD. F Floating bars display the dynamic range for each method; bars extend from the minimum to maximum values, and the line denotes the median. Significance by either two-way ANOVA (B, E) or ordinary one-way ANOVA (C, F), adjusted with Tukey’s multiple comparisons test. ns: P = 0.9929 (B), P = 0.9642 (E); **P < 0.01 (F, P = 0.0014): ***P < 0.001 (B, P = 0.0002; C, P = 0.0005; E, P = 0.0008). Source data is provided as a Source Data file. Created in BioRender. Srinivasan, R. (2025) https://BioRender.com/qkqi3qh. Development of an enhanced Dox-independent inducible Cas9. A Schematic diagram of the Xon based ultra-tight inducible Cas9. Destabilizing domain DD or DHFR was fused to the C-terminus of Cas9 to reduce the basal expression from the splicing switch. Alternatively, anti-CRISPR proteins with or without the FKBP degron were expressed from a human PGK promoter to inhibit Cas9 activity. The various Xon-Cas9 piggyBac constructs were stably transfected into HEK293T cells followed by drug treatment and measurement of CD81 levels as described in Fig. 1A and Table 1 (n = 3 biological replicates per stable line and treatment). B CD81 knockout efficiency was determined in the untreated (gray bars) and treated (colored bars) HEK293T cells. Bars show mean ± SD. C Floating bars show the dynamic range as the ratio of percentage CD81 disruption in treated vs. untreated cells for each method; bars extend from the minimum to maximum values, and the line denotes the median. D Time-course analysis of CD81 knockout in Xon-Cas9 vs. Xon-Cas9;AcrIIA4-FKBP HEK293T cells, that were left untreated or treated with either 20 nM Branaplam (Xon-Cas9) or 20 nM Branaplam + 1 μM dTagV-1 (Xon-Cas9;AcrIIA4-FKBP). Percentage CD81 negative cells were quantified by flow cytometry at the indicated time points (n = 3 biological replicates per stable line, treatment and time point). Data represent mean ± SEM. E, F K562 cells were stably integrated with the different versions of Xon based inducible Cas9 (n = 3 biological replicates per stable line and treatment). Bar graphs display CD81 perturbation efficiency in untreated (gray bars) and treated (colored bars, 150 nM Branaplam + 1 μM dTagV-1) cells. Bars show mean ± SD. F Floating bars display the dynamic range for each method; bars extend from the minimum to maximum values, and the line denotes the median. Significance by either two-way ANOVA (B, E) or ordinary one-way ANOVA (C, F), adjusted with Tukey’s multiple comparisons test. ns: P = 0.9929 (B), P = 0.9642 (E); **P < 0.01 (F, P = 0.0014): ***P < 0.001 (B, P = 0.0002; C, P = 0.0005; E, P = 0.0008). Source data is provided as a Source Data file. Created in BioRender. Srinivasan, R. (2025) https://BioRender.com/qkqi3qh. To evaluate editing kinetics in the Xon-Cas9 and Xon-Cas9;AcrIIA4-FKBP systems, we performed a time-course analysis with both flow cytometry (Fig. 6D) and targeted amplicon sequencing (Supplementary Fig. 6E). In the absence of drug treatment, Xon-Cas9 cells exhibited progressive loss of CD81 and increasing indel formation, whereas Xon-Cas9;AcrIIA4-FKBP cells maintained minimal background editing throughout the time course. Upon drug induction, Xon-Cas9 cells showed time-dependent indel formation and depletion of CD81, plateauing by 240 h. Drug-treated Xon-Cas9;AcrIIA4-FKBP cells exhibited slower editing relative to Xon-Cas9, but comparable knockout levels by 144 h. These time course results confirm that Xon-Cas9;AcrIIA4-FKBP is a superior system for controlling Cas9 activity. We then tested the Xon-Cas9;AcrIIA4-FKBP combination in K562 cells. Although background editing was nearly absent in this cell line with the pTET system, the unmodified Xon-Cas9 system displayed ~19% leaky target disruption (Fig. 6E). Use of the Xon-Cas9;AcrIIA4-FKBP vector reduced leakiness in this cellular context, but it also resulted in poor induction of Cas9 activity (Fig. 6E, F). Within this vector design, AcrIIA4 expression is under the control of a PGK promoter (Fig. 6A). Separate studies have observed slightly higher or more stable transgene expression from the PGK promoter relative to other constitutive promoters, albeit with context-dependent results47,48. Therefore, we hypothesized that the PGK-driven levels of AcrIIA4 protein were too high, even with dTag-mediated degradation, to fully reverse inhibition of Cas9 function, and that weaker AcrIIA4 expression was needed. To this end, we evaluated two distinct strategies aimed at lowering the baseline expression of AcrIIA4 such that it would maintain suppression of Cas9 in the uninduced state but allow for more complete reversal of this effect when cultured with a dTag molecule. This included exchange of the PGK promoter for the generally less-expressive UBC promoter47,48 or construction of an extended, polycistronic transgene with AcrIIA4 expressed by way of an IRES downstream of the CMV-Xon-Cas9 cassette (Supplementary Fig. 6F). IRES elements are commonly used to express multiple genes simultaneously and have been reported to drive lower expression of the downstream gene compared to the upstream, cap-dependent translated segment49. Both background and induced editing were greatly reduced with the UBC promoter driving the anti-CRISPR module, suggesting that UBC is stronger than PGK in our tested K562 cell line (Fig. 6E). By contrast, IRES-AcrIIA4-FKBP led to significantly reduced leakiness with no compromise in induced cutting and displayed the best dynamic range of editing (Fig. 6E, F). Together, these results demonstrate that non-Dox-inducible Cas9 expression systems can enable high efficiency and low background editing, aided through co-expression of degradable anti-CRISPR proteins, but cell type-dependent variability necessitates empirical testing for optimal performance. An enhanced, inducible CRISPRi system for potent, tightly regulated gene silencing The use of Cas9-mediated dsDNA breaks for gene perturbation may not be suitable for certain applications50. Accordingly, CRISPR interference (CRISPRi) provides a complementary and powerful tool for loss-of-function purposes. Here, a nuclease-dead Cas9 (dCas9) is fused to a Krüppel-associated box (KRAB) repressor or various chromatin and DNA modifying domains to facilitate transcription silencing of an intended target gene through a promoter or enhancer-targeting sgRNA51,52. While dCas9 fused to the KOX1 KRAB domain has commonly been used for CRISPRi, ZIM3 KRAB has emerged as a more potent alternative53,54, in both constitutive and inducible vector contexts53,55,56. Recognizing the utility of inducible CRISPRi for targeting both coding and non-coding regions, we sought to identify key parameters that enable robust, population-wide gene silencing and minimal background through expression of dCas9-KRAB variants. We began by evaluating a foundational set of TET-inducible CRISPRi systems. For this, KRAB domains from either KOX1 (pTET-KOX1) or ZIM3 (pTET-ZIM3) were fused to dCas9 and this cassette was positioned downstream of a TRE3G promoter within a lentiviral backbone with a constitutively expressed blasticidin resistance and rtTA transgene (Fig. 7A). Lentiviral infection with this vector was used to generate stable, polyclonal DLD1 (colorectal cancer) cell lines. These lines were subsequently transduced with lentivirus expressing an sgRNA targeting the CD81 promoter. Each cell population was treated with Dox or left untreated for 7 days and then assessed for CD81 depletion. The pTET-KOX1 cell line exhibited minimal effect on CD81 levels in the absence of Dox, but also correspondingly low (<50%) activity upon induction (Fig. 7B, Supplementary Fig. 7). Conversely, pTET-ZIM3 demonstrated substantial improvement for silencing CD81 upon Dox induction, although this potency translated to noticeable leakiness, impacting ~40% of the untreated population (Fig. 7B, Supplementary Fig. 7). This prompted us to optimize the ZIM3-based CRISPRi system to reduce its leakiness and maintain its activity via the principles applied for inducible expression of Cas9 nuclease.Fig. 7Ultra-tight inducible CRISPRi system.A Workflow for examining the inducible CRISPRi system in DLD1 cells. The inducible dCas9KRAB constructs were integrated into the cells via lentivirus transduction. The lentiviral cell line was established first and then the sgRNA was introduced in a second round of transduction. CD81 silencing in the inducible CRISPRi cells was measured after 7 days of treatment or non-treatment by flow cytometry (n = 3 biological replicates per stable line and treatment). pTET-KOX1, Tet-On dCas9 fused with KRAB domain from KOX1; pTET-ZIM3, Tet-On dCas9 fused with KRAB domain from ZIM3; pTET-DD-ZIM3, Tet-On system expressing a fusion protein containing DD, KRAB from ZIM3, and dCas9. B Bar graph displays the percentage of cells with CD81 silenced. Bars show mean ± SD. C Floating bars display CRISPRi dynamic range shown as the ratio of CD81 silencing in induced vs uninduced cells; bars extend from the minimum to maximum values, and the line denotes the median. D Workflow for examining PLK1 silencing in the inducible CRISPRi DLD1 cells. pTET-ZIM3 and pTET-DD-ZIM3 lines from (A) were transduced with lentivirus expressing sgRNA targeting PLK1 or a non-targeting control. The cells were selected and then treated or untreated for 5 days before measurement by Cell Titer Glo assay. E PLK1 silencing shown as normalized cell death. Normalized cell death is calculated as luminescence measurements of (cells with sgNTC—cells with sgPLK1)/(cells with sgNTC) in each condition. Bars show mean ± SD (n = 3 biological replicates per stable line and treatment). Significance by Two-way ANOVA with Tukey’s multiple comparisons test (B), uncorrected Fisher’s LSD (E), or RM One-way ANOVA (C). ns: P = 0.6174 (B), *P < 0.05 (C, P = 0.0183; E, P = 0.0106), **P < 0.01 (C, P = 0.0093), ****P < 0.0001 (B, E). Source data is provided as a Source Data file. Created in BioRender. Srinivasan, R. (2025) https://BioRender.com/tzgaun0. Ultra-tight inducible CRISPRi system. A Workflow for examining the inducible CRISPRi system in DLD1 cells. The inducible dCas9KRAB constructs were integrated into the cells via lentivirus transduction. The lentiviral cell line was established first and then the sgRNA was introduced in a second round of transduction. CD81 silencing in the inducible CRISPRi cells was measured after 7 days of treatment or non-treatment by flow cytometry (n = 3 biological replicates per stable line and treatment). pTET-KOX1, Tet-On dCas9 fused with KRAB domain from KOX1; pTET-ZIM3, Tet-On dCas9 fused with KRAB domain from ZIM3; pTET-DD-ZIM3, Tet-On system expressing a fusion protein containing DD, KRAB from ZIM3, and dCas9. B Bar graph displays the percentage of cells with CD81 silenced. Bars show mean ± SD. C Floating bars display CRISPRi dynamic range shown as the ratio of CD81 silencing in induced vs uninduced cells; bars extend from the minimum to maximum values, and the line denotes the median. D Workflow for examining PLK1 silencing in the inducible CRISPRi DLD1 cells. pTET-ZIM3 and pTET-DD-ZIM3 lines from (A) were transduced with lentivirus expressing sgRNA targeting PLK1 or a non-targeting control. The cells were selected and then treated or untreated for 5 days before measurement by Cell Titer Glo assay. E PLK1 silencing shown as normalized cell death. Normalized cell death is calculated as luminescence measurements of (cells with sgNTC—cells with sgPLK1)/(cells with sgNTC) in each condition. Bars show mean ± SD (n = 3 biological replicates per stable line and treatment). Significance by Two-way ANOVA with Tukey’s multiple comparisons test (B), uncorrected Fisher’s LSD (E), or RM One-way ANOVA (C). ns: P = 0.6174 (B), *P < 0.05 (C, P = 0.0183; E, P = 0.0106), **P < 0.01 (C, P = 0.0093), ****P < 0.0001 (B, E). Source data is provided as a Source Data file. Created in BioRender. Srinivasan, R. (2025) https://BioRender.com/tzgaun0. To improve dynamic range between the CRISPRi ON and OFF states, we incorporated an N-terminal DD tag to the ZIM3-dCas9 cassette, resulting in pTET-DD-ZIM3. In this instance, a separate anti-CRISPR module was not included to avoid complications with viral particle production related to cargo size limitations with lentiviral backbones. Induction of pTET-DD-ZIM3 cells by combined Dox and dTag treatment reduced CD81 levels similar to pTET-ZIM3 with minimal background gene silencing (Fig. 7B). As a result, dynamic range for the inducible, degron-enabled CRISPRi system increased ~13 fold over pTET-ZIM3 (Fig. 7C). Similar to our evaluation of pTET:Ultra-tight, we compared pTET-ZIM3 with pTET-DD-ZIM3 for disruption of the essential gene, PLK1 (Fig. 7D). As with pTET:Ultra-tight, background editing was largely eliminated with pTET-DD-ZIM3, and this coincided with a more robust phenotype upon induction (Fig. 7E). For CRISPRi, the combination of Dox induction with degron-mediated stabilization of ZIM3-dCas9 offers the most potent tool for gene silencing with minimal background. Discussion The ability to control gene expression timing and dosage is critical for probing dynamic and cell-state-dependent processes. Drug-inducible CRISPR/Cas9 systems address this need by enabling on-demand gene disruption under specific conditions that are otherwise masked or difficult to study with static approaches. These systems have already yielded meaningful insights, but their broader application has been tempered by background editing in the OFF-state and by only moderate efficiency in the presence of the inducer, relative to constitutive systems46. In particular, slow accumulation of DNA damage when inactive can degrade cell populations, complicating efforts to study essential genes3,4,14. To overcome these challenges, an array of chemically inducible CRISPR/Cas9 systems, such as split/intein Cas9 and ERT2-Cas9 have been developed, though these typically trade reduced OFF-state activity for diminished induction efficiency, with reported targeted rates of ~15%–60%5,6,10,11,13,57. To maximize the potency and limit background effects of conditional gene perturbation, we implemented complementary strategies for regulating Cas9 translation, stability, and function for development of the pTET:Ultra-tight system. To enable pTET:Ultra-tight for general use, we incorporated these modules into an all-in-one, stably integrating piggyBac vector. This vector allows for single-step creation of cell lines with the capacity for population-wide, inducible gene targeting and negligible baseline Cas9 nuclease activity without the need for single-cell cloning. Comparison of inducible Cas9 formats across cell lines (including iPSCs) and targets showed consistently high efficacy, low background, and superior overall performance with pTET:Ultra-tight. Furthermore, we demonstrated the ability of pTET:Ultra-tight to enhance the study of an essential gene (PLK1) by eliminating background editing, which can lead to pruning (death or reduced growth) of cells with leaky Cas9 activity. We also demonstrated a unique application for the pTET:Ultra-tight system via co-expression of target-flanked sgRNAs for controlled deletion of genomic loci. While we focused on the deletion of a protein-coding gene in this instance, this approach can likely be applied across a broad range of genomic elements, including unique exons or untranslated regions (UTRs), non-coding RNAs, and regulatory DNA. Multiplexed expression of sgRNAs (each directed to a distinct gene) from the pTET:Ultra-tight context should also facilitate targeting of multiple genes in the same cell. Evaluation of pTET:Ultra-tight across cell lines revealed areas for future system optimization. HCT116 cells demonstrated the lowest degree of target disruption (60-80%) in the presence of Dox, with or without the degron and anti-CRISPR control modules. This may be addressable through Dox-sensitized rtTA elements58. Conversely, while pTET:Ultra-tight showed the greatest reduction of background editing of all tested systems in SW480 cells, ~10% of cells showed uninduced target disruption. To reduce basal Cas9 activity, alternative degron modules or boosted expression of the anti-CRISPR element could prove beneficial. For either circumstance, different modules for induced Cas9 transcription may be more dynamic than the TET/Dox system, and these could be considered on a cell-type-by-cell-type basis for specialized applications59. Our results should also be considered in the context of other chemically inducible Cas9 strategies, such as split/intein-Cas9 and ERT2-Cas9 regulators5,6,10,11,13,57, which can reduce background activity but often at the expense of editing efficiency. By contrast, pTET:Ultra-tight consistently achieved 80–95% knockout with minimal leakiness across diverse cell types. For reference, Supplementary Table 7 summarizes reported efficiencies of representative split/intein Cas9 systems in comparison to the present work. We view these designs as complementary, with our system addressing a distinct need through the balance of near-zero background with strong induced performance in a broadly accessible format. Although pTET:Ultra-tight proved effective under a range of conditions, its reliance on Dox for induction makes it unsuitable for applications where this drug might impact cell physiology60. Separately, building complex synthetic gene circuits may require the use of Dox-induction for elements other than Cas9, necessitating alternative means of control for gene editing activities. To accommodate these possibilities, we took advantage of the recently developed Xon system, which enabled us to develop a module where Cas9 translation could be controlled through a Branaplam-induced alternative splicing mechanism. Despite strong Cas9 induction, resulting in population-wide editing in separate cell line contexts, removal of background editing required additional control measures. Comparison of degron-linked Cas9 to the degradable anti-CRISPR modules, where Cas9 would be induced by co-treatment with Branaplam and either Shield1 or a dTag molecule, showed lower uninduced editing with the anti-CRISPR module. Dosage of the anti-CRISPR was critical, and distinct expression strategies were required across cell lines to increase the ratio of Cas9 activity in the ON versus OFF states. Recently, the Xon system has been modified and used to reduce background editing of an inducible Cas9 cassette, albeit without additional modules61. We anticipate further evolution of this technology in the coming years, which will only increase the robustness and flexibility of inducible genome manipulation. However, Branaplam and other molecules in this class are expected to trigger alternative splicing of endogenously expressed transcripts62 beyond the SF3B3 intron used for the Xon module. This will need to be taken into consideration for select cell contexts and applications. Lastly, for experiments that benefit from targeted transcriptional repression and the avoidance of dsDNA breaks, we have established a lentiviral system for inducible CRISPRi. For this, we integrated the highly efficient ZIM3 KRAB domain within a dual Dox + degron-linked dCas9 expression context, resulting in pTET-DD-ZIM3. Combining these modules demonstrated superiority for both induced and background silencing levels compared to Dox + ZIM3-dCas9 alone or a Dox + dCas9-KOX1 KRAB fusion. As with pTET:Ultra-tight, the pTET-DD-ZIM3 provided substantially enhanced dynamic range and a more robust phenotype upon induction in a cell viability assay when targeting PLK1. Notably, gene silencing by the ZIM3 KRAB domain is expected to be transient. While spurious, low-level expression by Cas9 nuclease may result in measurable, progressive gene perturbation, this is less likely to be the case for ZIM3-dCas963. As such, adding a third control module within pTET-DD-ZIM3 (i.e., degradable anti-CRISPR) may offer little improvement in the dynamic range of ON vs. OFF-state silencing and could potentially reduce the quality of virus production or expression of transgenes when using lentivirus, given the limited cargo capacity of the lentiviral genome. Together, we provide an enhanced gene editing toolbox through the creation of an “Ultra-tight” inducible Cas9 nuclease, splice switch control over Cas9 activity, and related CRISPRi systems. These permit population-level gene disruption, often with marginal or no measurable background activity. The consistent and high dynamic range of efficacy for pTET:Ultra-tight was demonstrated in comparison to alternative inducible Cas9 formats across all tested cell types. We simplified and broadened the applicability of pTET:Ultra-tight by integrating it within an all-in-one transgenic vector for single-step cell line production and co-expression of one or more sgRNAs. Importantly, the core components and conceptual framework of our inducible gene perturbation systems, which rely on multi-layered control modules, could likely be adapted to create vectors for other functions, such as base editing, prime editing, or CRISPR activation, or to work with alternative effector proteins including Cas12a, Cas13, and their associated anti-CRISPRs20,64. With continuous innovation in targeted protein degradation65,66 and inducible genetic switch methodology, there will likely be more opportunities to expand the versatility and application of conditional genome engineering technologies67,68. While our study establishes the technical foundations of ultra-tight Cas9 regulation, several limitations should be acknowledged. We did not include characterization of novel phenotypes, nor did we perform rescue experiments post-disruption, which will be critical for dissecting or validating gene-specific mechanisms. Likewise, animal studies were beyond the scope of this work, although the regulatory modules used here (Tet/Dox, DD/Shield1, LID/Shield1, FKBP/dTAG, and Xon/Branaplam) have each shown in vivo functionality24,26,69–75 suggesting potential adaptation to animal models. In addition, while we summarized representative comparisons with split/intein-Cas9 and related chemically controlled systems, we did not perform head-to-head benchmarking in our own experiments. Together, these represent clear directions for extending this platform and point toward promising opportunities to apply ultra-tight Cas9 across both basic and translational settings, from probing essential gene function to enabling safer, more precise genome engineering in vivo. Methods Cell culture and stable line generation All parental cell lines, except KOLF2.1J, were obtained from Genentech cell bank (gCell), validated by STR profiling, and tested for mycoplasma. The KOLF2.1J iPSC line was obtained from the Jackson Laboratory (https://www.jax.org/jax-mice-and-services/ipsc/cells-collection/JIPSC001000; product code: JIPSC001000), maintained in feeder-free media, and validated by KaryoStat + ™(ThermoFisher Scientific), PluriTestTM (Thermo Fisher Scientific) and mycoplasma testing. Genome integration was accomplished using the PiggyBac transposition system where each piggyBac transposon plasmid was co-delivered into cells with a plasmid expressing the piggyBac transposase. Cell culture conditions and transfection methods are described in Supplementary Table 1. Briefly, stable integration in KOLF2.1J was achieved by nucleofection using a Lonza 4D-Nucleofector, while immortalized cell lines were seeded as described and transfected in 1 mL Opti-MEM, with the indicated transfection reagents as per manufacturer’s recommendations. Opti-MEM was replaced with growth media 4 h post-transfection. Plasmid constructs were generated by gene synthesis (Genscript) and included either Puromycin or Blasticidin resistance markers to select for stable cells. Unless otherwise indicated, stable lines were enriched with 1 μg/ml Puromycin or 20 μg/ml Blasticidin. Sequences of plasmids used in the study are listed in Supplementary Table 6. Guide RNA sequences for CRISPR/Cas9 nuclease-based editing and cell culture drugs used in this study are listed in Supplementary Tables 2 and 4. Small molecule dosing rationale All small molecules used in this study have been characterized in prior publications5,17,18,24–26,76–79 and were further titrated in our laboratory using luciferase or GFP reporter systems. Final working concentrations were chosen to maximize induction or degradation while minimizing cytotoxicity, and they consistently fall within ranges reported in the literature. Unless otherwise noted, experiments were conducted using the following concentrations: Doxycycline (2 µg/ml), Shield1 (1 µM), dTAGV-1 (1 µM), trimethoprim (2 µM), and Branaplam (20 nM in HEK293T and 150 nM in K562). Dose–response assays confirmed that these concentrations achieved near-maximal performance with minimal cytotoxicity. Flow cytometry In brief, adherent cells were lifted by TrypLE, or Accutase in the case of KOLF2.1J, at 37°C for 5 min and resuspended in growth media. Dissociated adherent cells or suspension cells were spun down at 500 g for 5 min and resuspended in 100 μl FACS buffer (PBS, 0.5% BSA, 0.05% NaN3 sodium azide) with 5 μL of antibody. After a 15-min incubation on ice in the dark, the cells were washed three times with FACS buffer and passed through a 35-μm cell strainer. Data was collected on BD Symphony instruments using FACSDiva acquisition software. Data was analyzed by FlowJo v10.10.0. The gating strategy to isolate live and single cell populations is shown in Supplementary Fig. 1B. The following antibodies were used for flow cytometric analysis. CD81 Monoclonal Antibody (1D6-CD81) APC (Thermo Fisher Scientific, Cat. No. 17-0819-42), mouse IgG1 kappa isotype control antibody APC (Thermo Fisher Scientific, Cat. No. 17-4714-82), SSEA-4 antibody Fluorescein (R&D Systems, Cat. No. FAB1435F), CD81 antibody PE (BioLegend, Cat. No. 349505), β2-microglobulin Antibody FITC (BioLegend, Cat. No. 316304), CD9 Antibody PE/Cyanine7 (BioLegend, Cat. No. 312115) and CD298 Antibody PE/Cyanine7 (BioLegend, Cat. No. 341707). Western immunoblotting Whole-cell extracts were prepared by lysing equal cell numbers in 1X Laemmli Sample Buffer (Bio-Rad, Cat. No. #1610737) for 10 min, followed by three cycles of sonication and boiling using a Bioruptor Plus (Diagenode). Proteins were separated on an SDS-PAGE and transferred to PVDF membranes using an iBlot2 Gel Transfer Device (Fisher Scientific, IB21001), followed by immunoblotting with the following antibodies: Cas9 (NovusBio, Cat. No. 7A9-3A3), α-Tubulin (Sigma Aldrich, Cat. No. CP06), Rabbit Anti-Mouse IgG(H + L)-HRP secondary antibody (SouthernBiotech, Cat. No. 6170-05), and Goat Anti-Rabbit IgG-HRP secondary antibody (SouthernBiotech, Cat. No. 4030-05). Chemiluminescence was performed with Amersham enhanced chemiluminescence (ECL) Prime reagent (Cytiva, RPN2232) and imaged on a ChemiDocTM MP Imaging System (Bio-Rad). Amplicon sequencing Genomic DNA was extracted from cells using the Agencourt DNAdvance Kit (Beckman Coulter). PCR primers flanking the target site were designed to amplify 200–300 bp amplicon using a 2 stage PCR method. Primers were synthesized at Integrated DNA Technologies (IDT). The PCR includes genomic DNA, i5 and i7 indexed adapters, target-specific primers with universal adapters, and PrimeTime Gene Expression Master Mix (IDT). The first stage of PCR amplifies the target-specific amplicons with target-specific primers flanked by universal adapters at 60°C Tm for 30 cycles. The second stage of PCR incorporates the indexed sequencing adapters in the target-specific amplicons at 55°C for 4 cycles. We use dual-end indexing by incorporating i5 and i7 10 bp indexes by IDT to uniquely label amplicons produced from each PCR. The PCRs are assembled using Echo 525 Acoustic Liquid Handler (Beckman Coulter). Following PCR, the amplicons from all of the reactions are pooled together. The amplicon library pool is cleaned twice using KAPA Pure Beads (Kapa) at a 1:1 ratio to remove unused primers and index adapters. The quality and quantity of the clean amplicon library pool is determined using 4200 TapeStation System (Agilent) and High Sensitivity D1000 DNA ScreenTape. The amplicon library pool is diluted to 15pM with 5% PhiX as per Illumina MiSeq library preparation procedure, and sequenced on Illumina MiSeq, dual read, 150/150 cycles to produce 300 bp sequences using the Miseq V2 kit (Illumina). Primer sequences used for amplicon sequencing are listed in Supplementary Table 5. Sequence analysis Sequencing reads were demultiplexed using i5 and i7 index barcodes, allowing up to one mismatch per index. Reads were then filtered to retain only those with a minimum Phred quality score of 30. For each reaction, reads were aligned to the expected amplicon sequence using BWA-MEM280. Alignment results were parsed using custom software to extract the CIGAR string (which encodes how each read aligns to the reference, including matches, insertions, and deletions). From this, the total number of aligned reads, the number of exact matches to the wild-type sequence, the number of reads with indels, and the number of partially aligned or non-matching reads were determined. Exact matches were classified as wild-type (unmodified), and reads with indels were classified as edited (reflecting sgRNA activity). Partially matched or non-specific reads were excluded from downstream analysis. Wild-type and indel-containing reads were then reported as a percentage of the total number of high-quality, specific alignments. Lentivirus production and transduction HEK293T cells were seeded and grown to 80-90% confluency at the time of transfection. Lentiviral plasmids expressing target genes were co-transfected with packaging plasmids pDelta8.9 and pVSVG into HEK293T cells using Lipofectamine 2000 reagent. 6 h after transfection, the medium containing the transfection mix was aspirated and fresh growth media was added to the cells. Lentiviruses were collected twice from the supernatant 24 h and 48 h post-transfection. Cell debris was removed by passing the crude supernatant through the 45 μm filter. Lentivirus was concentrated by Lenti-X (Takara Bio). DLD1 cells were cultured in RPMI 1640 supplemented with 10% TET-free FBS, 2mM L-Glutamine and pen-strep. Cells were seeded in 6-well Corning plates and grown to 50% confluency on the day of transduction. The lentivirus and 8 μg/ml polybrene were added to the DLD1 cells. Cells were spin infected at 1800 rpm for 45 min and incubated for 48 h before passage. DLD1 was first transduced with lentivirus expressing inducible CRISPRi systems and selected with 10 μg/ml Blasticidin to obtain the stable cell line, and then transduced with lentivirus expressing sgRNA and selected with 2 μg/ml puromycin. PCR genotyping Genomic DNA was isolated from untreated and treated cells 72 h post-treatment using QuickExtract™ DNA Extraction Solution (Lucigen). Both full-length and deleted forms of the target locus were amplified with Q5® High-Fidelity 2X Master Mix (NEB) using primers (forward 5′-GGCTTTGTAGAATGCTCGCTG-3′; reverse 5′-TTTAACGACCCCCAACTGCAT-3′) flanking exons 2 and 3 within the H3-3A gene. Primers were synthesized at IDT. Amplicons were resolved on an agarose gel and images were obtained using the ChemiDoc MP imaging system (Bio-Rad). Cell proliferation assay An equal number of HEK293T pTET-NTC, pTET or pTET:Ultra-tight stable cells expressing PLK1 sgRNA, were seeded at a low density in growth media with or without inducing drug(s). Cell growth was monitored using the IncuCyte S3 Live Cell Imaging and Analysis Instrument (Sartorius). Phase contrast cell images were collected using a 10× objective lens within the instrument every 4 h over 5 days and cell proliferation was determined based on confluence. Cell viability assay DLD1 stable cells generated by transduction of sgNTC lentivirus or sgPLK1 lentivirus cocktail (PLK1 sgRNA1 + PLK1 sgRNA2) were replated into 96 well format at 1000 cells/well in 80 μl medium. The following day, 20 μl of medium with or without treatment (final concentration: 250 ng/ml dox, 250 ng/ml dox + 1 μM shield1) was added to begin induction. On day 5 after induction, CellTiter-Glo Luminescent Cell Viability Assay (Promega #G7572) was used to read out cell viability on an EnVision plate reader (PerkinElmer/Revvity). Guide RNA sequences are provided in Supplementary Table 3. Figure production Flow cytometry plots and histograms were generated in FlowJo v10.10.0. Graphical representations (bar graphs, floating bars, and scatter plots) and statistical analyses were performed in GraphPad Prism v10. Illustration elements were created with BioRender.com. Statistics and reproducibility All experiments were conducted in the indicated cell lines under standardized culture and treatment conditions. Experimental groups (stable cell line, treatment, and time point) were defined by the study design. Sample sizes were established empirically based on reproducibility across independent experimental replicates, as specified in the figure legends, Methods, or associated text. For quantitative assays, independent biological replicates were performed to confirm reproducibility and enable statistical comparisons. Replicate numbers were chosen according to standard practice for cell-based and molecular assays. The exact n, statistical test, and data presentation format (e.g., mean ± SD/SEM or floating bars representing min–max with median line) are indicated in the figure legends. Statistical analyses were performed using GraphPad Prism v10. No statistical method was used to predetermine sample size. No data were excluded from analyses. Experiments were not randomized, and investigators were not blinded to sample allocation or outcome assessment. Materials availability Biological materials generated for these studies can be attained via MTA by contacting Andrew Ng (ng.andrew.an1@gene.com). Details related to the MTA process can be found on the Genentech Outgoing MTA Program website: https://www.gene.com/scientists/mta. Vector and guide RNA sequences are provided in Supplementary Tables. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Reporting Summary Transparent Peer review file Supplementary Information Reporting Summary Transparent Peer review file Source data Source data Source data