Epigenetic Engineering of K562 Cells: Dual-Vector Episomal Strategy for Stable Targeted DNA Methylation using dCas9-DNMT3A and -HDAC1 Fusion Proteins

Epigenetic regulation, including DNA methylation and histone modifications, plays a pivotal role in controlling gene expression, cell differentiation, and genome stability. Dysregulation of these mechanisms is often involved - either as a cause or a consequence - in various pathological states, particularly in cancer, neurological disorders, and imprinting syndromes^1,2,3. Notably, by precisely intervening to the epigenetic landscape at specific genomic loci using epigenome editing Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) technology, gene repression levels comparable to those observed in gene knock-out models can be achieved, without altering the underlying DNA sequence⁴. As a result, targeted epigenetic editing has emerged as a promising strategy for dissecting gene regulatory mechanisms and developing precision therapeutics.

DNA methyltransferase (DNMT) inhibitors and histone deacetylase (HDAC) inhibitors are epigenetic therapies currently approved for clinical use. However, these agents non-selectively target epigenetic markers across the genome, altering global chromatin states, causing off-target effects and toxicity in patients⁵. To address the limitations of traditional epigenetic alteration agents, programmable epigenetic editors based on catalytically dead Cas9 (dCas9) fused to effector domains, such as DNA methyltransferases (e.g., DNMT3A), demethylases (e.g., TET1), histone acetyltransferases (e.g., p300), or deacetylases (e.g., HDAC3), have been developed^6,7,8,9. The dCas9-effector system allows for locus-specific epigenome modification guided by a small RNA molecule, known as the single-guide RNA (sgRNA), without introducing double-stranded breaks at the target sequence. Recent studies have demonstrated their use in modifying gene expression, remodeling chromatin accessibility, and even reprogramming cellular phenotypes^10,11,12.

Despite rapid advances, dCas9-based epigenome engineering systems continue to face some limitations. The most common challenge is off-target effects due to the nonspecificity of epigenetic effector domains, especially with strong or prolonged expression of the dCas9 fusion tools^13,14. Additionally, the large size of dCas9 fusions presents significant delivery challenges. The size of the Streptococcus pyogenes Cas9 (SpCas9) sequence is approximately 4.2 kb¹⁵, which is comparable to the maximum capacity size of Adeno-associated virus (AAV) vectors. While lentiviral vectors have a relatively large capacity (~12-1 kb), enabling delivery of the full-length SpCas9 fused with epigenetic effectors, they still retain the possibility of random genomic integration, risking insertional mutagenesis and limiting their clinical applications¹⁶. On the other hand, non-viral approaches such as plasmid transfection suffer from poor retention in dividing cells. Transient or "hit-and-run" delivery approaches exclude the possibility of genomic integration, but often fail to sustain methylation changes across cell divisions, unless combined with chromatin remodelers or cooperative effectors^17,18,19.

To address these challenges, we employed a non-integrative, self-replicating episomal vector system (pEPI-S/MAR) for delivering dCas9-fused epigenetic effectors. This vector persists extrachromosomally in dividing cells at low copy numbers (~1-2 per cell), allowing for sustained yet moderate expression, in order to minimize undesired functions associated with overexpression^20,21. The Scaffold/Matrix Attachment Region (S/MAR) element of the episome interacts with the nuclear matrix, contributing to the episome's stable maintenance in the cells' nucleus during cell division^22,23. Furthermore, the pEPI vector system avoids risks associated with viral integration and enables delivery of constructs without exceeding packaging limits.

Recent studies have shown that promoter regions marked by active histone modifications, particularly H3K27 acetylation (H3K27ac), are often resistant to engineered DNA methylation. Combinatorial approaches that either remove or add histone marks using histone modifications' writers or erasers and simultaneously introduce DNA methylation via DNMT3A have been shown to increase the stability and persistence of gene repression, even in transient approaches^19,24. Based on these insights, we hypothesized that co-delivering dCas9-HDAC1 and dCas9-DNMT3A and targeting neighboring genetic areas would promote chromatin remodeling and facilitate more efficient and durable methylation at target regulatory elements.

In this study, we aimed to design and construct a system capable of delivering targeted and stable DNA methylation to CpG island 326 of ZBTB7A, a region characterized by low endogenous methylation and aberrant H3K27ac marks in K562 cells. The objective of this study was to deliver two episomal constructs encoding dCas9 fused to each of the distinct epigenetic effectors, along with sgRNAs, targeting the CpG-rich regulatory region upstream of the ZBTB7A gene using the low-expression, stable, non-integrative pEPI system. As illustrated in Figure 1, co-transfection of K562 cells with these episomes enables the dual recruitment of the editing complexes to the CpG island of interest, where HDAC1 removes local histone acetylation to facilitate subsequent DNA methylation by DNMT3A, resulting in coordinated and site-specific epigenetic modification.

The system described here is appropriate for applications where viral delivery is not feasible due to construct size, safety concerns, or cost limitations. It supports modular cloning of different sgRNAs and effector fusions, making it adaptable to a range of target loci. It is currently best suited to in vitro or preclinical studies in cell lines that tolerate plasmid transfection and antibiotic selection. In our implementation, episomal delivery in K562 cells was achieved using lipofectamine-based reagents and maintained under G-418 selection.

1. Design of sgRNAs

1. Identify the desired CpG island or regulatory CpG sites within the promoter region of the gene of interest using a genome browser (e.g., UCSC genome browser). Here, the design was applied on the CpG island 326 (GRCh38)of the ZBTB7A gene.
2. Design single guide RNAs (sgRNAs) using an online CRISPR sgRNA design platform [e.g., CHOPCHOP web tool (version 3)]. When designing sgRNAs, follow the recommendations below to ensure high on-target activity and minimal off-target effects:
   1. Select SpCas9 protein as the endonuclease, using 5'-NGG-3' as the protospacer adjacent motif at the 3' end (PAM-3').
   2. Choose sgRNA length to be 20 bp (besides the PAM sequence). Allow off-target matches with only three mismatches in the protospacer sequence.
   3. When selecting to target a specific region of the gene, choose the 'Promoter' option and define the upstream range (e.g., from 1500 bp to 3000 bp upstream for CpG island 326) to position the sgRNAs within the regulatory region.
   4. Use on-target efficiency scores such as those developed by Doench et al.²⁵, to assess sgRNA activity.
3. When selecting sgRNAs for experimental use, consider the following criteria:
   1. Rank all candidate sgRNAs by efficiency for the target region. Choose sgRNAs in the top 10-20% of predicted efficiency. When choosing, avoid sgRNAs with mismatches (MM0, MM1, MM2, MM3) at potential off-target sites.
   2. Choose sgRNAs that target originally unmethylated or weakly methylated areas for transcriptional repression using dCas9-DNMTs. For transcriptional repression using dCas9- HDACs, target highly acetylated regions for increased efficiency. Use online databases (e.g., ChIP-Atlas) to confirm DNA methylation and histone acetylation marks in K562 cells at the target loci.
   3. Use multiple sgRNAs to cover target regions larger than 100-200 bp²⁶.
      NOTE: Epigenetic modifications (methylation and deacetylation) influence regions upstream and downstream of the sgRNA binding site, and these effects might be limited to short genomic spans.
   4. Choose sgRNAs with a GC content between 40% and 80%.
      NOTE: The GC content of the sgRNA is important, as higher GC content improves RNA stability and dCas binding.
   5. Avoid sgRNAs containing stretches of four or more thymidines (Ts) and choose sgRNAs with adenine or thymine nucleotides located four bases upstream of the PAM site.
      NOTE: Four or more Ts can lead to premature transcription termination. sgRNAs with A or T located four bases upstream of the PAM site may exhibit stronger dCas binding.
   6. Check and avoid sgRNAs with high self-complementarity to minimize the risk of secondary structure formation.
   7. Finalize the selection of a few sgRNA sequences.
      NOTE: Four sgRNAs were designed spanning across the CpG island 326. The sgRNA1-4 target sequences and target positions across CpG island 326 are shown in Supplementary Table 1.
4. Transform the sgRNA sequence of choice (excluding the PAM triplet) into its reverse complement. Add GATCG to the 5' end of the forward sequence and G to its 3' end, and AAAAC to the 5' end of the reverse complementary sequence and C to its 3' end. Order the two final oligos for each sgRNA from a commercial oligo provider.

2. Cloning

NOTE: This section describes the tailoring of plasmid pEPI-1 to generate episomal vectors carrying both the dCas-epigenetic effector fusion and sgRNA expression cassettes. See Figure 2 for a comprehensive overview of the cloning workflow and vector architecture. All restriction digestion and ligation reactions described below were carried out according to the manufacturer's recommended protocols for each enzyme and reagent used, and reaction specifications are listed in detail in Supplementary Table 2.

1. Digest the pEPI-1 plasmid using AgeI and BspEI, which flank the eGFP (enhanced green fluorescent protein) coding sequence. Following restriction digestion, blunt the resulting overhangs using the Klenow fragment of DNA polymerase I. Ligate the blunt-ended plasmid backbone using T4 DNA ligase to generate the eGFP-deleted vector.
   NOTE: Deletion of the eGFP gene is performed to reduce the overall vector size. A larger vector size might limit the downstream cloning and transfection steps.
2. Obtain the coding sequences for the catalytic domain (CD) of the DNMT3A gene and the full-length of the HDAC1 gene. Append a stop codon at the 3' (downstream) end of each coding sequence. Add restriction sites BglII (5') and AseI (3') flanking the sequences.
3. Digest the eGFP-deleted pEPI-1 vector with BglII and AseI. Separately, digest the prepared DNMT3A(CD) and HDAC1 coding sequences with the same enzymes. Perform two independent ligation reactions:
   1. Ligate the DNMT3A(CD) fragment into the digested vector to generate the pEPI-1-DNMT3A construct.
   2. Ligate the HDAC1 fragment into the digested vector to generate the pEPI-1-HDAC1 construct.
      NOTE: BglII and AseI sites place the inserts downstream of the pCMV promoter and upstream of the S/MAR element for optimal expression.
4. Transform chemically competent Escherichia coli (e.g., NEB 5-alpha Competent E. coli, High Efficiency) with the ligation products from each vector separately. Plate the transformed cells on LB agar containing kanamycin (50 µg/mL) and incubate overnight at 37 °C. Select positive colonies and isolate plasmid DNA using the rapid boiling miniprep method^27,28.
5. Obtain the sequence encoding the catalytically dead SpCas9 (dCas9). Introduce the following elements in the specified order:
   1. At the 5' end: AscI restriction site, Kozak consensus sequence, and start codon (ATG).
   2. At the 3' end: NLS, optionally an HA-tag, followed by a BamHI restriction site.
6. Digest both pEPI-1-DNMT3A and pEPI-1-HDAC1 vectors with AscI and BamHI. Perform the same digestion on the dCas9 insert. Ligate the dCas9 fragment into each respective vector.
   NOTE: The AscI/BamHI sites ensure that dCas9 is inserted upstream of DNMT3A or HDAC1, directly downstream of the pCMV promoter. Ensure that a short linker or a few nucleotides are present between dCas9 and the epigenetic effector to maintain the correct reading frame.
7. Transform the final ligation products into high-efficiency competent E. coli strains with tight control of plasmid expression (e.g., NEB 5-alpha F'Iq Competent E. coli, High Efficiency). Select transformants on kanamycin plates. Prepare high-quality plasmid DNA from positive clones using the rapid boiling miniprep method.
8. Sequence the final plasmid constructs (pEPI-1-dCas9-DNMT3A and pEPI-1-dCas9-HDAC1) using nanopore sequencing to confirm the correct assembly of all components, verify the reading frame using an online translation tool (e.g., Expasy translate tool 3.0), and check for potential mutations or recombination events using an online alignment tool (e.g., BLAST alignment tool).
   NOTE: Large vectors are more prone to recombination and may accumulate mutations. Therefore, thorough sequence validation is essential.
9. Phosphorylate the 5' hydroxyl groups of both forward and reverse sgRNA oligonucleotides using T4 polynucleotide kinase. Hybridize the two complementary strands by heating the mixture to 90 °C for several minutes to denature secondary structures, then allow gradual cooling to room temperature overnight for complete annealing into double-stranded DNA.
10. Digest the selected pGuide sgRNA expression vector with BamHI and Esp3I. Ligate the annealed double-stranded sgRNA fragment into the linearized vector using T4 DNA ligase.
11. Using specific PCR primers (Supplementary Table 3), amplify the full gRNA expression cassette comprising (1) U6 promoter, (2) Cloned sgRNA sequence, and (3) sgRNA scaffold. Use a high-fidelity DNA polymerase that does not generate 3' overhangs (i.e., a proofreading polymerase) to ensure clean, blunt-ended PCR products suitable for downstream cloning.
12. Repeat steps 2.9-2.11 for each individual gRNA designed in section 1. Ensure each gRNA cassette is prepared and amplified separately to preserve target specificity.
13. Digest the vectors pEPI-1-dCas9-DNMT3A and pEPI-1-dCas9-HDAC1 with SacII. Blunt the sticky ends using the Klenow fragment of DNA polymerase I. Clone the previously amplified U6 promoter-sgRNA-scaffold cassettes into these vectors using T4 ligase.
    NOTE: Several gRNA-epigenetic effector combinations (e.g., sgRNA1-dCas9-DNMT3A, sgRNA2-dCas9-HDAC1, etc.) were constructed during the experimental phase. However, only two selected combinations were carried forward for the experiments presented in subsequent sections.
14. Transform the final ligation products into high-efficiency competent E. coli strains with tight control of plasmid expression. Select transformants on kanamycin plates. Prepare high-quality plasmid DNA from positive clones using the rapid boiling miniprep method.
15. Sequence the final plasmid constructs using nanopore sequencing, verify the reading frame, and check for potential mutations or recombination events.

3. Cell culture and transfection

1. Maintain K562 cells in T-75 culture flasks at 37 °C, under 5% CO₂ and humidity conditions. Use Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% Fetal Bovine Serum (FBS), 1× Penicillin/Streptomycin solution (Pen/Strep).
2. Prepare plasmids for transfection:
   1. Use 8 µg of total plasmid DNA per transfection, hence 4 µg of each plasmid for a co-transfection of two plasmids. Quantify the DNA using absorbance at 260 nm, and then dilute the required amount in 90 µL of ultra-pure water.
   2. Add 10 µL of 3 M Sodium Acetate solution and 2.5 volumes of 100% ethanol to the diluted DNA.
   3. Mix thoroughly and incubate at -20 °C overnight to precipitate the DNA.
      NOTE: When co-transfecting two plasmids with distinct gRNA-epigenetic effector combinations (e.g., gRNA1-dCas9-DNMT3A and gRNA2-dCas9-HDAC1), it is recommended to combine both plasmids into the same precipitation tube to simplify handling for downstream procedures.
   4. The following day, centrifuge the DNA precipitation tubes at 11,000 × g for 15 min at 4 °C.
      NOTE: Perform all subsequent steps under sterile conditions in a biosafety cabinet, as this stage will also involve handling live cells.
   5. Carefully discard the supernatant. Add 500 µL of 70 % ethanol to wash the pellet. Centrifuge at 11,000 × g for 5 min at 4 °C.
   6. Remove the ethanol supernatant completely. Allow the pellet to air-dry until no visible ethanol remains.
   7. Resuspend the dried DNA pellet thoroughly in 50 µL of DMEM. Keep the plasmid-DMEM solution at 4 °C until use.
   8. Run 400 ng of the prepared episomal DNA (approximately 2.5 µL of the resuspended solution) on a 1% agarose gel to verify the success of DNA purification and integrity prior to transfection.
3. Prepare cells:
   1. Harvest K562 cells from a T-75 flask when cultures reach 50-60% confluency to ensure transfection efficiency.
   2. Count cells using a hemocytometer. Transfer a volume containing 2 × 10⁵ cells into a 35-mm culture dish containing 900 µL of DMEM without FBS or antibiotics.
      NOTE: Use serum-free DMEM for all steps in the transfection procedure. Resume full medium culture conditions only after the transfection incubation period is complete.
   3. Place the dish in the incubator (37 °C, 5% CO₂) to maintain the cells ready for transfection.
4. Use a Lipofectamine-based transfection kit for plasmid delivery. Follow the manufacturer's instructions for the specific kit used, adjusting volumes if needed.
   1. Add 3 µL of P3000 Reagent (supplied in the transfection kit) to the prepared plasmid-DMEM solution. Mix gently and incubate the plasmid-P3000 reagent mix at room temperature for 5 min.
   2. Add 5 µL of Lipofectamine 3000 Reagent (supplied in the transfection kit) along with enough DMEM to reach a total reaction volume of 100 µL. Incubate at room temperature for 15 min.
   3. Add the full 100 µL transfection mixture to the 35-mm dish containing cells. Gently shake the plate to disperse the mixture, then return the dish to the incubator.
5. At 24 h post-transfection, add 3 mL of full medium (DMEM supplemented with 10% FBS and 1× Pen/Strep) to each 35-mm dish to initiate proliferation of transfected cells. At 72 hours post-transfection, begin the selection procedure by adding G-418 antibiotic to a final concentration of 1 mg/mL.
6. Once selection is adequately established and the total cell count exceeds 1 × 10⁶, proceed to isolate single clones (Figure 3A) using fluorescence-activated cell sorting (FACS):
   1. Harvest the cells by centrifugation at 300 × g for 6 min at 25 °C. Wash once with 1× sterile phosphate-buffered saline (PBS, pH ~7.4) and centrifuge again under the same conditions.
   2. Resuspend the cell pellet in 1× PBS and determine the cell concentration using a hemocytometer. Dilute the suspension to 1 × 10⁶ cells/mL, ensuring thorough resuspension.
   3. Set up FCS shorter to single-cell mode (ensures 1 cell per droplet), using a 96-well tissue culture-treated plate as the collection device. Adjust software settings if needed, depending on the instrument used (Figure 3B).
   4. Gate the primary population based on expected size (FSC) and granularity (SSC). Exclude debris (low FSC/SSC) and highly granular or dead material (high SSC). Use FSC-A vs. FSC-H or SSC-A vs. SSC-H dot plots to gate singlets.
   5. Pre-fill the 96-well plate with 100 µL of full medium containing 1 mg/mL G-418 per well. Warm the plate in the incubator prior to sorting.
   6. Perform the sort in single-cell mode, maintaining a sorting rate below 300 events/second to enhance sorting accuracy.
   7. After sorting, inspect the wells under a microscope to confirm one cell per well. Place the plate in a 37 °C incubator with 5% CO₂.
7. Allow single-cell clones to expand individually in the 96-well plate. Once clonal growth is established, reduce G-418 concentration to 400 µg/mL to maintain selection.
8. Transfer well-growing clones sequentially to 12-well plates, then to T-25 and finally T-75 flasks as needed, until each clone reaches ~1 × 10⁶ cells.
9. Collect double cell pellets at regular intervals. Store a pellet at -20 °C for genomic DNA extraction. Additionally, store a pellet homogenized in 500 µL of Trizol Reagent at -20 °C for RNA extraction and cDNA synthesis.
   CAUTION: Trizol Reagent should be handled with caution due to its toxicity and corrosive nature in a chemical fume hood while wearing personal protective equipment, and contaminated materials must be labeled for hazardous waste disposal.
10. Extract genomic DNA from selected clones using traditional phenol-chloroform extraction, followed by ethanol precipitation.
    CAUTION: Phenol:chloroform is toxic and corrosive, and should be handled in a chemical fume hood while wearing personal protective equipment. Dispose phenol- and chloroform-containing waste as hazardous chemical waste.
11. Ensure DNA is fully resuspended in nuclease-free water or TE buffer. Measure DNA concentration and purity using a spectrophotometer before proceeding to downstream applications.
    NOTE: This DNA will be used in downstream experiments such as PCR and Pyrosequencing to assess epigenetic editing efficiency.
12. Extract total RNA from cell pellets using Trizol Reagent, followed by isopropanol precipitation²⁹. Handle RNA samples with caution by ensuring RNase-free reagents and disposables, and maintaining samples at 4 °C during the procedure.
    NOTE: Total RNA will be used for cDNA synthesis using a reverse transcription kit.

4. PCR to confirm the presence of both episomes in clones

1. Harvest approximately 5 × 10⁵ cells from each culture clone, after resuspending cells thoroughly. Centrifuge at 7,000 x g for 3 min at 25 °C. Discard supernatant medium.
2. Wash the cell by resuspending in 1× PBS. Centrifuge again under the same conditions.
3. Discard the supernatant, leaving a small volume, just above the pellet level. Use 1 µL of the cell suspension directly as a template for PCR.
4. Alternatively, use 200 ng of purified genomic DNA extracted from the transfected clones.
   NOTE: Direct PCR from washed cells provides a rapid screening method but may yield variable results. For higher specificity and sensitivity, use purified genomic DNA as the template.
5. Prepare the PCR master-mix according to Table 1.
6. Use two sets of primers, with a common forward primer targeting the dCas9 sequence and separate reverse primers specific to each epigenetic effector vector (Supplementary Table 3).
7. Follow the cycling conditions presented in Table 2.
8. Run PCR products on agarose gel or analyze with capillary electrophoresis to verify product sizes: 230 bp for the DNMT3A-containing vector and 205 bp for the HDAC1-containing vector (Figure 3C). Include a positive control (mix of both plasmids at a concentration of 10-20 ng) to confirm the expected band sizes and validate PCR efficiency. Include a no template control (NTC).

Table 1: PCR reaction mix for episome verification.

Table 2: PCR cycling conditions for episome verification.

5. Targeted pyrosequencing analysis

1. Treat gDNA (1.5 µg) extracted from transfected clones with a bisulfite conversion kit according to the manufacturer's instructions.
2. Design primers flanking the CpG sites within the regulatory region targeted by the constructs. Use a dedicated bisulfite primer design tool. Follow these design guidelines:
   1. Convert the CpG island or CpG site regulatory region to the bisulfite converted version.
   2. Design a forward and reverse primer for PCR amplification of bisulfite-converted DNA. Include a biotin label on one primer (usually the reverse) for immobilization on streptavidin-coated beads.
   3. Design a sequencing primer (18-25 bp) internal to the PCR product, upstream of the first CpG of interest. As a target region, select a 100 bp window upstream or downstream of the sgRNA binding site to assess potential local epigenetic modifications.
   4. Avoid CpG sites within primer sequences to ensure unbiased amplification.
   5. Choose a primer pair with an Assay Design software score greater than 75 % and a quality rating of 'high' (blue), which indicates that the software analysis did not identify any problems or concerns.
      NOTE: For primer sequences designed to target CpG island 326, consult Supplementary Table 3. The primers were designed to bind in regions flanking the sgRNA1 and sgRNA2 binding sites, allowing sequencing of CpG sites surrounding the targeted loci.
3. Set up a PCR reaction using a high-fidelity hot-start DNA polymerase optimized for bisulfite templates.
4. Pyrosequencing procedure:
   1. Bind biotinylated PCR products to streptavidin-coated sepharose beads and isolate single-stranded DNA using a pyrosequencing workstation.
   2. Add the sequencing primer to the single-stranded DNA in the annealing buffer. Heat the mixture to 80 °C for 2 min, then allow it to cool gradually to room temperature.
   3. Perform sequencing reactions using the pyrosequencing instrument and the dispensation order defined during primer design.
5. Data analysis:
   1. Use dedicated pyrosequencing software to quantify the percentage methylation at each CpG site.
   2. Compare methylation levels between transfected and untransfected clones. Assess editing efficiency (Figure 3D). Consider performing a two-tailed unpaired t-test to compare methylation levels between untreated controls and each transfected clone at each individual CpG site.

6. cDNA synthesis and digital PCR for analysis of dCas expression

1. Prepare total RNA (1,100 ng per reaction) extracted from transfected clones in 12 µL of RNase-free water. Keep samples on ice.
2. Assemble the genomic DNA elimination mix on ice: Mix 2 µL of gDNA wipeout buffer, RNA sample containing 1,100 ng of total RNA, and RNase-free water to a final volume of 14 µL. Incubate for 2 min at 42 °C, then place on ice.
3. Prepare the reverse transcription master mix on ice: Mix 4 µL of RT Buffer with 1 µL of RT primers, and add 1 µL of reverse transcriptase.
4. Combine 14 µL of genomic DNA elimination mix with 6 µL of reverse transcription master mix to a total reaction volume of 20 µL. Mix gently, and incubate at 42 °C for 15 min. Inactivate the reverse transcriptase by heating at 95 °C for 3 min.
   NOTE: Store cDNA at -20 °C until further use.
5. Dilute cDNA from all samples 1:1 with nuclease-free water. Use the diluted cDNA for downstream digital PCR analysis, using a digital PCR system with dCas-specific primers (Supplementary Table 3):
   1. Prepare the 4× Master mix and Reaction mix according to Table 3. Add diluted cDNA to the mix and pipette gently to ensure thorough mixing.
      NOTE: Set up reactions in at least duplicates for all clones. Include a negative control (cDNA from untreated K562 cells) and an NTC.
   2. Load 9 µL of the Reaction mix into each well of the PCR plate. Set up the PCR system using the cycling conditions presented in Table 4.
   3. Select the FAM dye channel for detection and start the run.
   4. After completion, retrieve the samples and analyze the data.
      NOTE: Digital PCR reports results as copies per µL of reaction.
   5. Multiply the reported value (copies/µL) by the reaction volume (10 µL) to obtain total copies per reaction. Divide this value by the amount of cDNA in the reaction to obtain copies per ng cDNA (equivalent to ng RNA input).
      NOTE: If the exact protocol is followed, each sample begins with 1,100 ng of RNA in a 20 µL RT reaction (55 ng/µL). After 1:1 dilution, the working concentration is 27.5 ng/µL, and since 1 µL is added to each PCR reaction, this corresponds to 27.5 ng of cDNA (RNA-equivalent) per reaction.

Table 3: 4× Master mix and reaction mix for digital PCR analysis of dCas expression.

Table 4: Cycling conditions for digital PCR.

To investigate whether our targeted epigenetic editing tools could induce site-specific DNA methylation changes at the ZBTB7A CpG island 326, we first generated episomal vectors encoding dCas9 fused to either DNMT3A(CD) or HDAC1, along with various sgRNAs designed with target sites within CpG island 326. We initially co-transfected K562 cells with combinations of these plasmids to evaluate whether simultaneous delivery of sgRNA and effector modules could lead to reproducible epigenetic modifications.

Initial episomal co-transfection strategy and limitations
We first attempted to co-transfect four episomal plasmids simultaneously. Τwo vectors expressing dCas9-fused epigenetic effectors (pEPI-1-dCas9-DNMT3A and pEPI-1-dCas9-HDAC1) and two separate pEPI-1 plasmids carrying only the sgRNA expression cassettes targeting the CpG island 326 were initially used (Figure 2). Different combinations of sgRNAs (1-4) were tested, co-delivering two sgRNAs per transfection to cover regulatory sites across the CpG island 326. Target sites of sgRNAs across the CpG island 326 are illustrated in Figure 3D.

Despite successful transfection, pyrosequencing analysis revealed no consistent changes in DNA methylation near target sites of the sgRNAs employed. We hypothesized that this lack of detectable modification may be due to the episomal nature of the vectors. Since episomal vectors are maintained at a low copy number (~1-2 copies per cell) and a single plasmid suffices to confer G-418 resistance, it is likely that not all cells initially received or retained after selection the full combination of four plasmids. This limitation likely impaired the effectiveness of targeted epigenetic editing and the consistency of results across the population of the transfected cells.

Refined strategy: sgRNA integration into effector plasmids
To overcome the low retention of multiple plasmids, we prepared single constructs both encoding the dCas9-effector fusion protein and expressing a sgRNA. Specifically, we generated pEPI-dCas9-DNMT3A-sgRNA1 and pEPI-dCas9-HDAC1-sgRNA2 (Figure 2), targeting two neighboring regulatory CpG regions within the CpG island 326. These sgRNAs were selected based on several criteria. First, they had proximal binding sites flanking a 367 bp segment of the ZBTB7A CpG island. Additionally, their targeting region exhibited high levels of histone acetylation and low DNA methylation in K562 cells, indicating transcriptional activity and accessibility. Finally, sgRNA1 and sgRNA2, when transfected separately from the effector domain, occasionally induced weak methylation changes. This was inconsistent between biological replicates, possibly due to loss or unequal distribution of episomal DNA. By embedding the sgRNAs within the epigenetic effector vector, we ensured that each transfected cell would co-express both components.

To ensure that downstream analysis was performed only on cells carrying both constructs, we proceeded to sort single cells into 96-well plates using FACS (Figure 3A,B). After proliferation to a sufficient number of cells, we collected cell pellets from the surviving clones (n = 10) to verify the presence of both episomal constructs. Since each clone originated from a single cell, the episomal profile of each population reflected that of the initially transfected cell. This approach allowed us to selectively exclude any clones that did not contain both plasmids and retain the remaining positive clones (n = 3) for further analysis. The presence of both constructs was confirmed by a multiplex PCR using a common forward primer targeting the dCas9 sequence, and two reverse primers specific to either DNMT3A or HDAC1. A representative agarose gel showing both positive and negative clones is presented in Figure 3C.

Importantly, we continued to verify the presence of both episomes in the positive clones by PCR for up to 40 days in culture, confirming stable maintenance of the constructs over time. In addition to confirming episome presence by PCR, we quantified the dCas-system expression in double-positive clones. On average, clones containing both constructs displayed 103.2 ± 16.6 copies per ng RNA of dCas expression, confirming the transcriptional activity of the episomal system.

Clone selection and preliminary methylation results
Following clonal selection, pyrosequencing of clones that carry both plasmids revealed that a portion of the transfected population exhibited slight but statistically significant increases in methylation at CpG sites located near the gRNA1 target region. All three positive clones showed elevated methylation levels compared to untransfected K562 controls, with no significant differences observed between them. Figure 3D presents a representative histogram from clone C2, showing the percentage of methylation across individual CpG sites upstream of the sgRNA1 binding site, 40 days after initial transfection. In contrast, the assay flanking the sgRNA2 target site showed no significant increase in methylation.

Figure 1: Schematic overview of the dual CRISPR-dCas9 epigenetic editing system. K562 cells were co-transfected with two episomal vectors, one encoding dCas9 fused to the catalytic domain of DNA methyltransferase DNMT3A and sgRNA1, and another encoding dCas9 fused to the histone deacetylase HDAC1 and sgRNA2. The pEPI vector persists extrachromosomally via an S/MAR element and possesses the ability to replicate in dividing cells without integrating into their genome. Upon episomal entry and expression, the dCas9 fusion proteins are directed to the nucleus and then target adjacent sequences within the ZBTB7A CpG island by their respective sgRNAs. HDAC1 deacetylates local histones and DNMT3A methylates nearby CpG sites. This combinatorial strategy is designed to promote stable and locus-specific epigenetic repression. Please click here to view a larger version of this figure.

Figure 2: Construction of episomal CRISPR-dCas9 effector plasmids. The base vector pEPI-1, containing S/MAR, SV40 origin, and eGFP reporter, was first modified by deletion of the eGFP cassette to create an intermediate cloning vector. Subsequently, epigenetic modifiers -DNMT3A (CD) or -HDAC1 were inserted, followed by cloning of dCas9 to generate pEPI-1-dCas-DNMT3A and pEPI-1-dCas-HDAC1 constructs. Individual single-guide RNAs (sgRNA1 or sgRNA2), driven by the U6 promoter, were then integrated into these effector plasmids to produce pEPI-1-dCas-DNMT3A-sgRNA1 and pEPI-1-dCas-HDAC1-sgRNA2. Final plasmids retain essential features for episomal maintenance and antibiotic selection, including the S/MAR sequence, Neomycin/Kanamycin resistance gene, and SV40 polyadenylation and replication elements. Co-transfection of these plasmids into K562 cells is recommended over the initial strategy that employed separate vectors for sgRNAs and epigenetic enzymes, as described in the representative results section. S/MAR: scaffold/matrix attachment region; NLS: nuclear localization signal; U6: human U6 promoter; CD: catalytic domain; pCMV: cytomegalovirus immediate-early promoter; sgRNA: single guide RNA and scaffold; SV40: Simian Virus 40 polyadenylation and replication elements; eGFP: enhanced green fluorescent protein. Please click here to view a larger version of this figure.

Figure 3: Targeted epigenome editing of CpG island 326 using CRISPR-dCas9 constructs in K562 cells. (A) Schematic overview of the workflow: K562 cells were transfected with episomal constructs encoding dCas9-DNMT3A and dCas9-HDAC1 fusions. Single cells were sorted into 96-well plates using a flow cytometer and clonally expanded. Positive clones were screened by PCR and analyzed via pyrosequencing. (B) Screenshot of the cell sorter software, showing single cell sort mode in a 96-well format. (C) PCR confirmation of dual construct presence in clones using agarose gel (3 %) electrophoresis. Bands at 230 bp and 205 bp correspond to dCas9-DNMT3A and dCas9-HDAC1, respectively. Lane 1: clone positive for pEPI-1-dCas-DNMT3A-sgRNA1; Lane 2: clone positive for pEPI-1-dCas-HDAC1-sgRNA2; Lane 3: clone positive for both constructs; Lane 4: positive control (plasmid DNA mix); NTC: no-template control; Lane L: 100 bp DNA Ladder. (D) Pyrosequencing analysis of CpG methylation across seven CpG sites in CpG island 326. Clone C2 (green) shows increased methylation compared to untreated K562 cells (blue) at multiple sites (* = p < 0.05). Error bars indicate standard deviation. sgRNA target positions relative to the 367 bp analyzed region are mapped below the bar graph. Please click here to view a larger version of this figure.

Supplementary Table 1: Target sequences of sgRNAs 1-4 designed to direct dCas9 constructs towards CpG island 326 of ZBTB7A and their genomic location. Please click here to download this File.

Supplementary Table 2: Recommended reaction volumes, reagent amounts, and typical conditions for cloning steps. Please click here to download this File.

Supplementary Table 3: Primer sequences for cloning, PCR, and pyrosequencing assays that are described in the protocol section. Please click here to download this File.

One of the most critical steps in this protocol is the design of highly specific sgRNAs. Off-target effects are a known limitation of CRISPR-based epigenetic editing systems, which can often arise from dCas9 binding infidelity, as well as the non-specific activity of the effector domains. To minimize this, sgRNAs must be designed using bioinformatic tools that optimize for high on-target activity and minimal off-target binding³⁰. Employing design tools and validating predicted off-target sites is recommended, while comparing clones carrying different sgRNAs can further help discriminate true on-target effects. Another key step is the use of only the catalytic domains of the epigenetic modifiers used, to prevent endogenous interactions by the retained function of the full-length enzyme³¹. It is important to note that the effectiveness of dCas-fused systems is susceptible to several factors that are not always predictable, such as the chromatin state, the loci surrounding the target area within a 3D structure context in the nucleus, and the cell type. Selecting sgRNA target sites is often approached experimentally in a trial-and-error way, which is time-consuming. Designing multiple sgRNAs spanning the target CpG region and prioritizing loci with low endogenous methylation and/or high acetylation for HDAC targeting increases the likelihood of obtaining effective clones. Online tools predicting sgRNA activity and the effect of CRISPR-Cas9 epigenetic editing systems are also becoming available^32,33,34.

Transfection efficiency and episome retention are important determinants of the protocol's success. The lipofection-based transfection of K562 cells yielded viable clones in approximately 10 out of 96 sorted wells, each stably carrying both episomal constructs. This recovery rate is modest, although it was sufficient to identify and expand clones for downstream analysis. To enhance outcomes, plasmid integrity should be verified with agarose gel electrophoresis, and endotoxin-free, high-purity DNA should be used. Including a mock-transfection control (without addition of episomal DNA) is recommended to monitor baseline cell viability. During clonal expansion, maintaining G-418 at a stabilizing concentration and delaying its reduction until adequate growth is achieved helps preserve episome retention. Re-verifying episome presence by PCR at multiple passages is essential to confirm long-term maintenance.

Single-cell sorting is another critical stage where troubleshooting is often required. Survival rates can be improved by pre-filling wells with conditioned medium containing G-418 and 10-20 % FBS, reducing the sorting rate (<300 events/sec) to minimize droplet stress, and delaying the lowering of G-418 concentration until clones are visibly expanding. Early PCR screening, conducted prior to clonal expansion, helped prioritize wells with positive clones and avoid unnecessary downstream processing.

Episomal delivery helps minimize the risks associated with genomic integration and off-target effects. However, the low number of dCas-effector molecules retained per cell using this approach may come at the cost of reduced methylation efficiency. It has been shown that lowering the expression levels of dCas9 methyltransferase and sgRNAs can indeed reduce off-target methylation, but this often negatively affects the on-target efficiency of the system¹⁴. In our case, the stable and long-term expression achieved through episomal maintenance, combined with the use of two epigenetic effectors capable of reinforcing epigenetic memory, was essential to compensate for the low-copy imbalance.

Another limitation is the need for selection and clonal expansion to identify cells that retain and express both constructs, which can be a laborious process. Nonetheless, it is only when selecting single cells that a significant outcome is observed in contrast to the heterogeneous cell population, prior to sorting.

Finally, this protocol is optimized for cell lines, and its application in primary or hard-to-transfect cells might be less efficient. Adaptation of this system for in vivo use would require further engineering to enhance episome delivery and clone selection. The current design employs G-418 selection and clonal expansion to maintain dual-plasmid retention and ensure long-term stability. This strategy is well tolerated in K562 cells and suitable for proof-of-concept studies; however, it is less suitable for primary cells that are more sensitive to selective pressure. In such cases, fluorescent reporter-, incorporated in the plasmid backbone, based approaches (e.g., eGFP or mCherry), could offer a faster and less toxic alternative for identifying positive cells, minimizing the need for single-cell cloning and repeated PCR verification. Besides, the strength of S/MAR based episomal vectors has been previously tested in human hematopoietic progenitor cells^35,36.

Compared to viral systems such as lentiviruses, our method offers a non-integrative alternative that supports long-term, low-level expression of epigenetic effectors. While transient systems provide short-term expression, they typically fail to produce stable methylation unless combined with multiple effectors or chromatin remodeling components, as mentioned. The dual-effector approach used here (HDAC1 + DNMT3A) builds on recent findings suggesting that H3K27ac is a key feature associated with resisting methylation-directed silencing and locus transition to inactive chromatin¹⁹. By removing acetylation through HDAC1, we assist DNMT3A access and methylation efficiency, supporting more persistent silencing.

To our knowledge, no previous studies have reported the delivery of Cas proteins using S/MAR-based episomal vectors. Episomal systems have been employed in other contexts for the delivery of CRISPR tools, particularly for difficult-to-transfect cells such as pluripotent stem cells, but they typically rely on alternative strategies such as OriP/EBNA1-based vectors^35,36,37.

S/MAR retention and transgene expression in K562 cells have been validated in previous studies^20,21,38. S/MAR elements support episomal replication and long-term maintenance by interacting with transcription "factories" within the cell nucleus, and preserving chromatin accessibility³⁷. These elements, when embedded in episomal vectors, prevent epigenetic silencing, enhance transgene expression, and ensure stable propagation at low copy numbers without selection^21,23,38. These vectors co-segregate with chromosomes during mitosis and offer a high cloning capacity³⁹, making them a competitive alternative to integrative viral systems.

Finally, it is important to note that when CRISPR-Cas tools are overexpressed, off-target binding of Cas proteins has been observed, and reducing plasmid copy number and controlling expression levels can mitigate these effects^40,41. The episomal S/MAR system, which maintains vectors at low copy numbers per cell and ensures stable expression, may thus offer an advantage in reducing unintended interactions compared to high-expression integrative approaches.

While this study establishes a reproducible protocol for targeted methylation in K562 cells, additional improvements are required before episomal tools can be applied in therapeutic contexts. Safety concerns remain before therapeutic use can be envisioned. These include the risk of off-target modifications inherent to CRISPR-based tools and pre-existing immunity to Cas proteins in human populations. Both issues represent significant hurdles that are currently being actively investigated through improved enzyme engineering, optimized delivery methods, and comprehensive preclinical testing^14,42,43.

Looking forward, targeted epigenome editing tools can provide mechanistic insight into epigenetic regulation in various disorders and develop therapeutic approaches, as epigenetic alterations are increasingly being explored as contributing factors in imprinting disorders, certain cancers, and neurological conditions. In addition, dCas-epigenetic tools might be used to generate model animals that mimic pathological phenotypes, offering insight into epigenetic dysregulation without altering the DNA sequence. While further work is needed to optimize delivery to primary tissues and organisms, non-integrative systems like pEPI-1 offer a safer starting point for preclinical applications.