Plasmid2MC: efficient cell-free generation of high-purity minicircle DNA for genome editing in mammalian cells Authors: Roman Teo Oliynyk, Ahmed Mahas, Emil Karpinski, George M Church Published: 2025 ABSTRACT AbstractDNA plasmids are widely used for delivering proteins and RNA in genome editing. However, their bacterial components can lead to inactivation, cell toxicity, and reduced efficiency compared to minicircle DNA (mcDNA), which lacks such bacterial sequences. Existing commercial kits that recombine plasmids into mcDNA within proprietary bacterial strains are labor-intensive, yield inconsistent results, and often produce endotoxin-contaminated low-quality mcDNA. To address this challenge, we developed Plasmid2MC, a novel cell-free method utilizing ΦC31 integrase-mediated recombination to efficiently excise the bacterial backbone from conventionally prepared plasmids, followed by digestion of the bacterial backbone and all other DNA contaminants, resulting in highly pure and virtually endotoxin-free mcDNA. We demonstrated the application of mcDNA to express CRISPR-dCas9 for base editing in HEK293T cells and mouse embryonic stem cells, as well as for homology-independent targeted insertion (HITI) genome editing. The method’s ease of preparation, high efficiency, and the high purity of the resulting mcDNA make Plasmid2MC a valuable tool for applications requiring bacterial backbone-free circular DNA. Subject terms: DNA recombination, Genetics research, Single-molecule biophysics, Enzymes Introduction Plasmids, bacterial extrachromosomal self-replicating circular DNA elements, were first described by Joshua Lederberg in 19521. They were later found to facilitate the sharing of antibiotic resistance and other inter-bacterial genome modification2,3. Cohen et al.4 demonstrated the construction of biologically functional bacterial plasmids in vitro in 1973. Since then, plasmids have been widely used for genome editing constructs delivery5–7, including zinc-finger nucleases, transcription activator-like effector nucleases (TALENs), and CRISPR-based methods8–12. However, the prokaryotic sequences within plasmids, notably antibiotic resistance genes, are redundant for eukaryotic cell transfection and have the potential to elicit adverse effects on gene expression, cellular physiology, immune response13–15, and lead to plasmid inactivation15. Minicircle DNA (mcDNA), produced from plasmids by eliminating bacterial backbones, can avoid these effects. mcDNA has shown lower inactivation, better nuclear uptake, and greater protein expression compared to plasmids15–17, making it a promising tool for non-viral vector delivery18. Numerous studies have demonstrated the production of mcDNA using site-specific recombination13,14,19–21, where a serine integrase performs double-stranded DNA cleavage and subunit rotation to exchange DNA strands at two recombination sites present in a plasmid—the so-called attP and attB (abbreviated from attachment-phage and attachment-bacteria, respectively) sites22—thus separating the intended mcDNA and a bacterial backbone into two intercoiled DNA circles (see illustration in the Results). Similar approach was also used with Cre (tyrosine) recombinase, however due to Cre recombinase operating as reversible equilibrium, all being equal this approach will lead to lower yield than ΦC31 integrase23–26 The recombination is then followed by a restriction enzyme digestion. Both recombinase and restriction enzyme genes are designed into specialized bacterial strains and an inducible activation is used to express them, resulting in the production of mcDNA within the bacteria13,14,20,21; with exception of Mayrhofer et al., who instead of restriction enzyme digestion performed a specially designed chromatographic matrix purification step25. We found the commercial kit27 using this methodology labor-intensive and yielding inconsistent results and excessive mcDNA endotoxin levels. Purity and toxicity are known issues for such bacterially recombined mcDNAs, and additional cleanup is also cumbersome and labor-intensive18. In a recent analysis, Almeida et al.18 highlighted a common issue with specialized bacterial kits utilizing ΦC31 integrase and I-SceI: incomplete recombination and digestion, resulting in mcDNA tainted by remnants of parental plasmids and recombined empty bacterial backbone plasmids, which are unacceptable for therapeutic applications, posing significant challenges for downstream purification28–30. No protocol or commercial kit for cell-free plasmid recombination into mcDNA for use in genome editing experiments has been available. To address this shortcoming, we present Plasmid2MC, a cell-free protocol leveraging ΦC31 integrase-one of the best-characterized large serine recombinases to date31–37. In Plasmid2MC, ΦC31 recombination efficiently excises the bacterial backbone from conventionally prepared plasmids in a cell-free reaction, while spin column mcDNA extraction steps simultaneously reduce endotoxin levels, thus not requiring additional purification. The ease of preparation, efficiency, and mcDNA purity of Plasmid2MC make it a superior alternative to existing kits on the market, such as those implementing inducible recombination and plasmid digestion within specially modified bacterial strains, produced by System Biosciences Inc. (SBI)-currently the only company offering such commercial kits. Our evaluation of SBI’s MC-Easy™ Minicircle DNA Production Kit revealed a costly, time- and labor-intensive process prone to multiple failure points and yielding mcDNA with high endotoxin levels and DNA contamination. Although recombination using ΦC31 integrase and other recombinases has been studied in laboratory experiments for decades, the Plasmid2MC method represents a novel, first-of-its-kind published protocol that delivers a simple and efficient cell-free approach for the practical production of high-quality mcDNA. Finally, we demonstrated the efficacy of mcDNA prepared with the Plasmid2MC method by using it to express CRISPR-dCas9 for base editing in HEK293T and mouse embryonic stem cells (mESCs), as well as for homology-independent targeted insertion (HITI) genome editing. Results Method overview In this study, we developed an efficient cell-free method that takes as an input customarily cloned plasmids containing two recombination sites (attP and attB, Fig. 1a), and with the help of ΦC31 integrase recombines these plasmids into two intercoiled DNA circles, one containing a bacterial backbone with all components used for bacterial amplification, and another containing the mcDNA of interest Fig. 1b, c. ΦC31 recombination: mechanism diagram and experiment. a Diagram of a plasmid containing attP and attB recombination sites separating the bacterial backbone and the intended mcDNA. b Subunit rotation mechanism for strand exchange by serine recombinases (this subplot is adopted from Rutherford et al. 201376). c Random collision and topologically selective strand exchange both lead to intercoiled structures of two DNA loops of varying complexity among the recombined molecules22. Random collision is the usual recombination mechanism, while topologically selective recombination was observed in experimental designs22. d Gel image showing the ΦC31 recombination of 10418 bp plasmid into a 2.7 kbp bacterial backbone circle and 7758 bp DNA circle. Each lane contains 300 ng of DNA. Lane M - DNA ladder; L1 - pure supercoiled plasmid; L2 - same plasmid with a single cut with a restriction enzyme (NotI); linear DNA travels faster than supercoiled DNA, and L1 presents a very weak matching band showing a minuscule amount of linear plasmid; L3 - after 80 minutes of recombination showing diffuse DNA scattering corresponding to the varying speed of travel of the miscellaneous catemers depicted in (b), and L4 same product cut with NotI, which is present in a single location on the bacterial backbone, thus allowing bacterial backbone release from the tangle, resulting in distinct bands (2.7 and 7.7 kbp) and a weaker band corresponding to the non-recombined plasmid; L7 and L8 - same as L3 and L4 after 12 hours of recombination; the original plasmid band is now nearly invisible; L9 and L10 - purified mcDNA following digestion of all non-circular DNA (see the unprocessed gel photos in Supplementary Fig. S1). In the past, we developed a fully synthetic one-pot method for producing small circular DNA (up to 1000 bp) that uses only one spin column cleanup step (see our publication Oliynyk & Church38). We tried to apply the same one-cleanup approach to the Plasmid2MC method to avoid the inevitable loss of mcDNA during cleanup. However, preliminary experiments have shown that the cleanup step after the completion of recombination was indispensable in Plasmid2MC, even though it lowered the final mcDNA yield by estimated 16% for 5.7 kbp source plasmid size and by 31.5% for 13.9 kbp source plasmid (see Methods DNA purification after the completion of recombination and Protocol steps). The recombination buffer containing 20 mM Tris, 0.1 mM EDTA, and 100 mM NaCl worked optimally (see Fine-tuning recombination buffer composition for optimal performance in the Methods), and experiments were performed with a recombination duration of 12 h (see the finalized Protocol steps in the Methods section). The recombination reaction was then followed by a 1 h Proteinase K digestion to remove the ΦC31 integrase, succeeded by the first QIAGEN QIAquick spin column cleanup of the recombined mcDNA, which was still intertwined with the bacterial backbone plasmid. QIAGEN advises using Proteinase K digestion for DNA samples with high protein content when using their DNA purificationspin column kits, which are highly selective against proteins39. This step indeed resulted in high-purity DNA with 260/280 ratios of 1.85 and 260/230 ratios of 2.17 for a 10.4 kbp source plasmid (see Recombination Yield Table.xlsx in Supplementary Data). The DNA recovery was on average 83.6% (a DNA loss of 16.4%) during the spin column cleanup for this plasmid size. Subsequently, we cleaved the empty bacterial backbone plasmid using the SspI restriction enzyme and degraded all linear and nicked DNA with T5 exonuclease, retaining only the intended mcDNA. This was followed by a second QIAquick spin column cleanup, yielding further purified mcDNA (see Supplementary Data and Methods section for detailed procedures). Plasmid2MC protocol yields at varying ΦC31 integrase and source plasmid DNA concentrations The combined recombination results are presented in Fig. 2 (see also Component concentrations in recombination experiments in the Methods section). Three plasmid sizes were used in the following analysis, titled as mcDNA size/source plasmid size in base pairs (bp)-3096/5758bp, 7758/10418bp, and 11282/13942bp-thus sampling the recombination protocol performance across a commonly used range of plasmid sizes and corresponding mcDNAs (see Plasmid design and assembly in the Methods, Supplementary Data, and Supplementary Fig. S2). In Fig. 2a, b, both ΦC31 integrase and source plasmid DNA (10418 bp plasmid used) had concentrations of 100 ng/μl. Recombination yield gradually increases between 18 and 30 °C, reaching 38% at 30 °C, which is the optimal temperature (see Fig. 2a). Recombination yield diminishes thereafter to 5% at 37 °C. The reaction yield increased with longer recombination times, even showing additional gains from 33.3% at 4 hours to 39.4% at 8 hours of incubation, and remaining at a constant level thereafter (see Fig. 2b). Recombination reaction efficiency and mcDNA yields. a Temperature curve of ΦC31 recombination efficiency, showing the distinct maximum yield at 30 °C. b Recombination yield by time at the optimal temperature of 30 °C. c mcDNA yield at ΦC31 integrase concentrations of 25–200 ng/μl and a constant DNA concentration of 100 ng/μl. d mcDNA yield at 100 ng/μl ΦC31 and DNA concentrations of 100–300 ng/μl. e mcDNA yield at 200 ng/μl ΦC31 and DNA concentrations of 100–300 ng/μl. The following plots (f–h) account for the bacterial backbone DNA circle for complete recombination efficiency (including length of bacterial backbone, and discounting processing losses): f Recombination yield at ΦC31 concentrations of 25–200 ng/μl and a constant DNA concentration of 100 ng/μl. g Recombination yield at 100 ng/μl ΦC31 and DNA concentrations of 100–300 ng/μl. h Recombination yield at 200 ng/μl ΦC31 and DNA concentrations of 100–300 ng/μl. i Combining the data points shown above presented a molar ratio of ΦC31 to source plasmid DNA molecules. j Maximum yield test at ΦC31 concentration of 200 ng/μl and 20 μg of input DNA at a concentration of 66.6 ng/μl. The column titled 1211/3871bp represents yields for an additional 1211 bp mcDNA produced out of 3871 bp parental plasmid. Note: Tests for each data point were performed in duplicate for all subplots, with exception of (j) performed in quadruplicate and triplicate, as seen in the subplot. Experiments in plots (b–j) were performed at 30 °C and 12 h incubation. Further investigations with the three test plasmids at a constant plasmid DNA concentration of 100 ng/μl and variable ΦC31 integrase concentration (see Fig. 2c) showed that at 25 ng/μl of ΦC31, mcDNA yield was low for all three plasmids, gradually increasing with increasing concentrations, then beginning plateauing at 100 ng/μl. An additional investigation with fixed ΦC31 integrase at 100 ng/μl (see Fig. 2d) showed a slight yield decrease for 10,418 and 13,942 bp plasmids when DNA concentration increased to 200 ng/μl, with a steeper drop at 300 ng/μl. At this ΦC31 integrase concentration, the yield for the shortest 5758 bp plasmid nearly halved from 27.2 to 17.2% when DNA concentration doubled; when DNA concentration tripled to 300 ng/μl, the yield decreased to 7.6%. This suggests that a 100 ng/μl ΦC31 integrase concentration may have been insufficient for the recombination of the short plasmid at higher DNA concentrations. Increasing the ΦC31 integrase concentration to 200 ng/μl improved performance by 1–2% for all three source plasmids at a DNA concentration of 100 ng/μl and mitigated the performance loss for the 5758 bp plasmid at higher DNA concentrations, resulting in parallel performance patterns for all three plasmids (see Fig. 2e). This indicated that smaller plasmids, with a larger relative number of attP and attB sites per unit of weight, may require higher concentrations of ΦC31 integrase. Interestingly, the mcDNA yield was similar (25–30%) for the three plasmids at the most effective concentrations. While the above results report the pure mcDNA yield, which are the values we measured directly, the successful recombination reaction includes a circular bacterial backbone of constant size, which was cut with a restriction enzyme and digested. Thus, the full recombination efficiency can be recalculated as: where Reff is the full recombination efficiency, Myield is the measured mcDNA yield, and Psize and Msize are the sizes of the plasmid and mcDNA, respectively. Figure 2f–h are equivalent to Fig. 2c–e, showing the full recombination efficiency while accounting for the digested bacterial backbone according to the formula above. Recombination performances at concentrations of 200 ng/μl of ΦC31 integrase and 100 ng/μl DNA for the aforementioned experiments were 54%, 40%, and 32% for the shortest, medium, and longest plasmids, respectively. One explanation for this performance difference is that the cleanup column efficiency decreases with increasing DNA length, with column losses from two cleanups exceeding 50% for the largest source plasmid (see Protocol steps, Step 2). Additional DNA nicking or shearing may occur during initial plasmid extraction or intermediate cleanup, and these molecules will be digested by T5 exonuclease during the digestion stage. However, despite higher losses during cleanup for larger plasmids, the proportion occupied by the bacterial backbone in the larger plasmids was smaller than, for instance, in the 5758 bp plasmid, where the bacterial backbone made up nearly half of the plasmid size. This accounts for the similar mcDNA yield, despite the more significant differences in full recombination efficiency. Recalculating the molar concentration of ΦC31 integrase and dividing it by the molar concentrations of the three test plasmids from the available data points (see Fig. 2i, and Component concentrations in recombination experiments in the Methods) demonstrated that the recombination reaction yield is relatively low when ΦC31 molar excess is at or below 25X, and begins plateauing for all three plasmids at approximately 50X ΦC31 molar excess, with only fractional improvement at higher molar excesses. This figure is a sampling of clearly non-linear correspondence between concentrations and the molar excesses of these two variables. For example, at higher concentrations (DNA at 300 ng/μl and ΦC31 integrase at 200 ng/μl), the relative performance is notably below the trend line for lower molar ratios (see circled red data points for 10,418 bp and 13,942 bp plasmids in Fig. 2i), showing that at such a high DNA concentration, the efficiency of recombination is diminishing. The experiments described above were performed with elution volumes of 22–25 μl, which aimed to achieve higher concentrations of DNA, thereby allowing for accurate measurements. Based on the observations in Fig. 2i, we decided to verify the maximum yields at a concentration of ΦC31 integrase equal to 200 ng/μl, with a reaction volume of 300 μl and double the source plasmid weight of 20 μg (DNA concentration of 66 ng/μl), using an elution volume of 54 μl. After incubating the columns at 55 °C for 3 minutes before elution, centrifugation yielded 48–51 μl of mcDNA eluate. The results presented in Fig. 2j indicate that even with a higher elution volume, the recombination yield remained essentially identical to the results for ΦC31 integrase concentrations of 200 ng/μl in Fig. 2h. The 13,942 bp source plasmid showed a 1.4% improvement, while the 10,418 bp source plasmid exhibited a yield lower by 0.7%–within the inter-experimental variability. In this scenario, only the smallest 5758 bp source plasmid demonstrated a 3% improvement, resulting in a maximum recombination efficiency of 56.8% and an mcDNA yield of 30.5% at this plasmid size. This improvement was likely due to the molar ratio of ΦC31 integrase to DNA increasing from 87 in the first data point in Fig. 2h to 130 for this source plasmid size. This outcome aligns with our experience that heating the QIAquick cleanup columns at 55 °C for 3–4 minutes before centrifugation results in an effective QIAquick column elution rate when using elution volumes of 20–25 μl, with a repeated elution from the same column yielding only a few additional ng/μl. This shows that the use of lower elution volumes, necessary for achieving high mcDNA concentration required for genome editing experiments, will not lead to significant mcDNA losses with QIAquick spin columns. A question arose regarding the Plasmid2MC method’s capability to produce even smaller mcDNAs. To address this, we developed a shorter version of the plasmid, p-att-SuperShort1211bp, which measures 3871 bp and contains the same bacterial backbone, yielding mcDNA with a length of 1211 bp (see Methods). The yields of this construct are shown in the first column of Fig. 2j, labeled 1211/3871bp (also see the sequencing diagram in Supplementary Fig. S3). As anticipated, due to the smaller ratio of mcDNA size to the total plasmid size, the yield of mcDNA was relatively lower compared to longer constructs, at 15.7%. However, the overall recombination efficiency, accounting for all processing losses, remained high at 50.1%. The molar ratio for the reaction at the concentrations listed in Fig. 2i was 88. It is expected that with an even smaller mcDNA-to-source-plasmid ratio, the mcDNA yield would decrease proportionally; nevertheless, smaller circular constructs can be produced more efficiently with our CV protocol, as further discussed in the Discussion section. The high molar ratios of ΦC31 integrase raise a pertinent question: if only four molecules are sufficient for recombination per source plasmid, why do our experiments demand such an excess of integrase? Beyond the canonical ΦC31 attP and attB sequences (which do not exist verbatim in the human genome), approximately 27,924 pseudo-sites within the human genome can mimic these attP and attB functions40. These pseudo-sites facilitate the formation of functional ΦC31 dimers roughly every 107 kbp, a pattern potentially applicable to any DNA segment or plasmid. Moreover, the number of sites where a single ΦC31 molecule can bind, whether transiently or persistently, is significantly higher, contingent upon the binding energy of each site41,42. Such off-target binding may limit the availability of exactly four ΦC31 molecules that need to simultaneously combine - two at attP and two at attB - for recombination to take place. In Supplementary Fig. S21, we reanalyzed data from Fig. 2I to estimate the number of ΦC31 molecules per kbp of source plasmid. The occurrence of these single-integrase binding sites varies with plasmid DNA sequence, and indeed Supplementary Fig. S21 indicates that recombination efficiency plateaus when exceeding five ΦC31 molecules per kbp, with all tested plasmids showing diminished recombination below 2.5 ΦC31 molecules per kbp. Optimal concentrations of source DNA and ΦC31 integrase Having consolidated the above data, we can formulate the considerations for the optimal concentrations and quantities of ΦC31 integrase and source plasmids. Plasmid Midipreps typically yield 400–600 μg of plasmid per 45ml of E. coli culture, providing abundant source plasmid DNA. To economize the use of ΦC31 integrase, it may be optimal to use 100 ng/ul of ΦC31 and 200–300 ng/μl of DNA for larger plasmids (e.g., see Fig. 2d, g) and a somewhat higher ΦC31 concentration for smaller plasmids (see the 5758 bp plasmid in Fig. 2e, h). While it is practical to measure concentrations in ng/μl, it is important to keep in mind that the molar ratio (or recalculate in less common units of ΦC31 molecules per kbp of source plasmid) is a crucial factor for determining the quantities used in Plasmid2MC reactions. Plasmid2MC method’s concomitant reduction of endotoxin levels in mcDNA compared to source plasmids Although no specialized cleanup steps were employed beyond the two spin column purifications required for Plasmid2MC processing, we observed a significant reduction in endotoxin levels. This finding warranted a more detailed investigation, described below. Tests using the Endosafe® nexgen-MCS endotoxin testing equipment revealed that source plasmids prepared with the QIAGEN Midiprep Plus kit had endotoxin levels ranging from 0.17 to 1.46 EU/μg of DNA. While the QIAGEN kit specifies endotoxin levels below 1 EU/μg, one of three samples slightly exceeded this limit (see endotoxin testing reports in Supplementary Fig. S20 and preliminary data in Supplementary Fig. S15). To assess a scenario with slightly elevated initial endotoxins, we selected a parental plasmid with a level of 1.46 EU/μg of DNA and determined that the Plasmid2MC protocol lowered the endotoxin content in the resulting mcDNA to 0.042 EU/μg-a 35-fold reduction (see Fig. 3a). Notably, although the protocol incorporates two spin column cleanups with washing steps, these alone do not suffice for substantial endotoxin removal. Reduction of endotoxin levels during plasmid2MC processing. a A 35-fold reduction in endotoxin concentration was observed from the source plasmid to the resultant mcDNA (see endotoxin testing reports in Supplementary Fig. S16). b Digestion experiments supported the hypothesis that proteins involved in digestion might bind endotoxins, facilitating their removal during the Plasmid2MC spin column purification step (see endotoxin testing reports in Supplementary Fig. S17). We hypothesized that endotoxin reduction occurs during the digestion stage, where T5 exonuclease and the SspI restriction enzyme bind endotoxins, forming protein-endotoxin complexes43,44 that are subsequently removed during washing. To test this, we used the plasmid p-SBkit-ABE8e-parental-plasmid.gbk, which lacks a recognition site for the SspI restriction enzyme (see Fig. 3b and corresponding endotoxin testing reports in Supplementary Fig. S17). Incubating 10 μg of this plasmid per step 3 of the Plasmid2MC protocol reduced endotoxin levels 4-fold, from 0.170 EU/μg to 0.039 EU/μg after cleanup. However, introducing Proteinase K after the digestion step resulted in no improvement of endotoxin levels, which remained close to the initial level. Thus, using Proteinase K after the recombination reaction is crucial (see Materials and methods), but proteinase digestion should not follow T5 exonuclease and restriction enzyme digestion. Comparative evaluation of the system biosciences Inc. MC-Easy™ minicircle DNA production kit While of secondary importance for the purposes of this research, we performed a detailed comparative analysis of the commercially available SBI MC-Easy™ kit, presented in the Supplementary Chapter titled ‘Comparative evaluation of System Biosciences Inc MC-Easy™ Minicircle DNA Production Kit,’ with key results summarized in Supplementary Figs. S11–S14, S18. Testing showed that the recommended procedure produced mcDNA with endotoxin levels ranging from 47 to 121 EU/μg, accompanied by detectable traces of parental plasmid, recombined empty bacterial backbone, and E. coli genomic DNA (see Supplementary Figs. S11–S12). While the optional DNase digestion and subsequent cleanup steps recommended by SBI effectively eliminated DNA contamination, they also led to a significant loss of mcDNA. After this additional purification, endotoxin levels in mcDNA generated by the SBI MC-Easy™ Kit were reduced by only approximately 3-fold, remaining high at 14.2–45.8 EU/μg. The SBI MC-Easy™ kit is expensive (see pricing comparison in the Supplementary and in Discussion section) and demands a 3-4-day preparation and quality analysis process, featuring at least four possible major failure points-if any fail, the entire minicircle preparation must be restarted. The limitations of SBI MC-Easy™ approach (see Supplementary Information for details) likely arise from recombining and digesting plasmids within living E. coli cultures-a method proven too complex for consistent efficacy and prone to significant endotoxin contamination due to the large volumes (400 ml) of bacterial culture required, which ultimately yield relatively low amounts of mcDNA. See Table 1 for comparison of Plasmid2MC and MC-Easy™ methods. Table 1Comparison of Plasmid2MC vs MC-Easy™ kit for equivalent yields of mcDNAExtractedAfter purificationMethodInitial inputmcDNAEndotoxinmcDNAEndotoxinCostPlasmid2MCE.coli ≈ 10 ml190 μg21.46 EU/μg225 μg0.042 EU/μg$75.703(SD)(0.11)(1.24)(0.05)MC-Easy™E.coli 400 ml4105 μg584.7 EU/μg525 μg630.0 EU/μg6$225.20(SD)(6.4)(41.6)(13.6)(17.4)The values in this table are aggregated from Fig. 2, Fig. 3, Supplementary Fig. 14, and the Supplementary chapter dedicated to MC-Easy™ kit evaluation and the comparative cost estimates. The two method steps differ, the specifics for each method are listed in this footnote. 1About 1/5th of the 45–50 ml of E.coli culture processed by Midiprep. 2For Plasmid2MC this is about 1/5th plasmid DNA (pDNA) extracted using a Midiprep. Showing pDNA exact amount used for recombination and endotoxin level. There was 0% of bacterial DNA present after Midiprep. 3The ΦC31 integrase cost included in this total was $7.37. 4The MC-Easy™ bacterial culture volume after arabinose induction; it may be increased to 600 ml total at higher optical density (OD). 5The MC-Easy™ mcDNA yield and endotoxin level for mcDNA after Maxiprep extraction. The DNA contained on average 5.5% of bacterial DNA, and contaminations with the 4.2 kbp MC-Easy™ recombined empty bacterial backbone plasmid and 11 kbp parental plasmid, requiring MC-Easy™ DNase digestion and purification step that included alcohol precipitation. 6The final MC-Easy™ yield and endotoxin levels after the purification step. (SD) denotes the value’s standard deviation, where applicable. Comparison of Plasmid2MC vs MC-Easy™ kit for equivalent yields of mcDNA The values in this table are aggregated from Fig. 2, Fig. 3, Supplementary Fig. 14, and the Supplementary chapter dedicated to MC-Easy™ kit evaluation and the comparative cost estimates. The two method steps differ, the specifics for each method are listed in this footnote. 1About 1/5th of the 45–50 ml of E.coli culture processed by Midiprep. 2For Plasmid2MC this is about 1/5th plasmid DNA (pDNA) extracted using a Midiprep. Showing pDNA exact amount used for recombination and endotoxin level. There was 0% of bacterial DNA present after Midiprep. 3The ΦC31 integrase cost included in this total was $7.37. 4The MC-Easy™ bacterial culture volume after arabinose induction; it may be increased to 600 ml total at higher optical density (OD). 5The MC-Easy™ mcDNA yield and endotoxin level for mcDNA after Maxiprep extraction. The DNA contained on average 5.5% of bacterial DNA, and contaminations with the 4.2 kbp MC-Easy™ recombined empty bacterial backbone plasmid and 11 kbp parental plasmid, requiring MC-Easy™ DNase digestion and purification step that included alcohol precipitation. 6The final MC-Easy™ yield and endotoxin levels after the purification step. (SD) denotes the value’s standard deviation, where applicable. In comparison, the Plasmid2MC method employs routine bacterial amplification of the source plasmid. Subsequent steps, outlined in the Methods section, are conducted entirely in a cell-free, in vitro system, enabling precise timing in a thermocycler. As a result, Plasmid2MC is faster, more cost-effective, and more reliable, producing notably higher-quality mcDNA than the MC-Easy™ Kit. With endotoxin levels as low as 0.042 EU/μg, Plasmid2MC-derived mcDNA is exceptionally well-suited for transfection into sensitive cell lines. The mcDNA use in CRISPR-dCas9 base editing in HEK293T and mESCs Having optimized the protocol using ΦC31 integrase for the effective production of mcDNA, it is customary to demonstrate the performance of the resulting DNA constructs. For this validation, we used the plasmid p-att-ef1a-ABE8e-dCas9-BSD (see the Materials and methods) and corresponding mcDNA produced using our protocol. This plasmid is an implementation of TadA for the base editing version ABE8e introduced by Richter et al.11, assembled into our plasmid containing ef1α promoter, attP and attB sites, blasticidin selection resistance gene, and dCas9—which is used in our ongoing research (rather than nCas9 used by Richter et al.11; see Fig. 4a). Applying the Plasmid2MC protocol to the source plasmid p-att-ef1a-ABE8e-dCas9-BSD, we produced mcDNA for use in these validation experiments. The resulting mcDNA was primarily represented by a single circular DNA, confirming the known ΦC31 integrase preference for intra-molecular recombination22, with a small proportion of double-sized intermolecular concatemers, marked “dimer" on the sequencing histogram from Plasmidsaurus (see Fig. 4b, Supplementary Fig. S2). The sequencing showed the perfect matching of the expected and achieved mcDNA (sequencing files are available in the Supplementary Data). Validation of mcDNA performance in ABE8e dCas9 base editing. a Diagram of the source plasmid and the resulting mcDNA with excluded bacterial backbone, used in this validation edit. b Plasmidsaurus sequencing histogram showing the high quality of the mcDNA product. c A circular vector for the expression of gRNA for base editing targeting, using the method developed earlier in our laboratory38. d Base editing (dCas9 ABE8e) A → G conversion rates in HEK293T cells for the first three sites from11. e dCas9 ABE8e editing A → G conversion rates in two mESC sites. Colors: mcDNA (green) and the original plasmid (orange). Guide sequences for the sites are followed by PAM (in bold red font). In plots (d, e), standard deviation (SD) values are indicated with the red line range. Where not visible, the SDs fall within the spread of the data points. We used the Circular Vector (CV) protocol to prepare the five gRNA expression vectors used for five base editing experiments (see Supplementary Fig. S4). Briefly, the CV protocol takes a linear double-stranded DNA fragment and ligates/purifies it into short, circularized DNA containing the gRNA expression element, as illustrated in Fig. 4c, which can be completed and ready for transfection in as little as 3 hours (for implementation details, see Oliynyk & Church38). This protocol is particularly useful for making multiple gRNA expression vectors with minimal hands-on time, especially when used in conjunction with larger plasmids expressing proteins (and now mcDNAs), which are more labor-intensive but can be reused for multiple editing targets. We targeted three sites from Richter et al.11, demonstrating A → G conversion in HEK293T cells (see Fig. 4d). Figure 4e also demonstrates successful A → G conversion at two locations in mESCs using the same plasmid and mcDNA. The editing of both HEK293T and mESCs was followed by antibiotic (blasticidin) selection. The editing performance was similar in magnitude and differed only slightly between the plasmid and the mcDNA. The editing efficiency for HEK293T cells was consistently high, due to our extensive experience with this cell line, and additional attempts at optimizing transfection and selection parameters resulted in minimal improvement. However, after several trials, we succeeded in optimizing the transfection and selection of mESCs (see Methods), achieving an editing efficiency above 80%. We performed NGS sequencing to validate these results, and indeed, the editing efficiency was even slightly higher than what was observed in EditR Sanger sequencing analysis in Fig. 4e, showing 86.5% editing for position A6 and 81.7% editing for position A10 for mESC Site 4 (see Supplementary Fig. S7). These experiments showed comparable base editing performance when using the plasmid and mcDNA in both cell types, as should have been anticipated in such basic validation experiments. Maniar et al.16 demonstrated minicircle inactivation in non-dividing liver cells starting four days after transfection, with increasing inactivation thereafter. By this time, the dilution in concentrations of both mcDNA and plasmids in HEK293T cells, and even more so in mESCs, would far outweigh the eventual inactivation of expression vectors in terms of base editing efficiency. The mcDNA use in homology-independent targeted integration (HITI) mcDNAs have garnered significant interest due to their utility in site-specific genome integration applications. HITI utilizes mcDNA for the insertion of exogenous DNA into targeted genomic sites without the need for homologous recombination45,46. Here, we further demonstrated the utility of the Plasmid2MC method to produce HITI-compatible mcDNAs. We designed mcDNA constructs encoding a fluorescent protein cassette (mKate) linked to an antibiotic selection marker (Hygromycin) via a T2A peptide sequence, which allows for the in-frame insertion of both the fluorescent protein and the selection marker into housekeeping genes, including ACTB and GAPDH (see Fig. 5a). To showcase the preparation of larger HITI templates, we added a non-functional random DNA sequence downstream of the hygromycin resistance gene, increasing the total size of the HITI template to approximately 3 kbp. The integration sites were strategically chosen just upstream of the stop codons of these housekeeping genes. A P2A sequence was incorporated immediately upstream of the donor template sequence to ensure that the insertion of the HITI constructs occurred in-frame, facilitating the translation of the downstream HITI elements as part of the same polypeptide as the endogenous proteins (see Fig. 5a). We employed the CRISPR-Cas9 system with gRNA CVs specifically designed to target these integration sites alongside our mcDNA constructs. Following transfection, cells were cultured under hygromycin selection for two weeks to ensure only those cells that successfully integrated the mcDNA constructs survived (see Fig. 5b). HITI editing results. a Schematic showing the design of HITI constructs. Blue pentagon, Cas9/gRNA target sequence in the host genome and HITI mcDNA. HygR: Hygromycin-resistant gene. Black square: non-functional random DNA sequence. b Schematic of HITI experiment with antibiotic selection to enrich cells with the correct HITI-mediated integration before subsequent analysis. c Flow cytometry analysis of cells treated with HITI or control (untransfected cells). Percentage of positive population is shown within the gates. d Gel electrophoresis of PCR overlapping the mKate gene and either the ACTB or GAPDH genes. HITI - shows bands for edited samples, while no bands are visible for unedited Controls. See Supplementary Fig. S9 for additional analysis and unprocessed gel electrophoresis photos. Flow cytometry analysis revealed a significant population of cells expressing mKate, with 91.6% of ACTB-targeted cells and 62.2% of GAPDH-targeted cells displaying the red fluorescent protein, indicating successful integration and expression (see Fig. 5c). To confirm the integration of the HITI mcDNA at the genomic level, PCR was conducted using primers specific to the mKate gene and either the ACTB or GAPDH genes. The expected amplicons were observed only in HITI-treated cells, with no amplification in control groups (see Fig. 5d). Subsequent Sanger sequencing of these amplicons verified the precise integration of the HITI constructs at the targeted genomic loci, confirming the correct junctions between the mcDNA and the host genome. These findings substantiate the efficacy of the Plasmid2MC method for producing large HITI-compatible mcDNA donor templates. Discussion The Plasmid2MC method and protocol presented here elucidate the preparation of mcDNA from conventionally amplified or cloned plasmids containing ΦC31 attP and attB sites at the junctions between the bacterial backbone and the intended mcDNA, facilitated by ΦC31 integrase. In this protocol, ΦC31-mediated recombination efficiently excises the bacterial backbone, followed by purification steps that digest the separated bacterial backbone and reduce endotoxin levels (see Materials and methods). Our approach provides an alternative to costly proprietary kits that rely on specialized bacterial strains, yielding highly purified mcDNA with low levels of bacterial endotoxins. Specifically, the SBI MC-Easy™ kit is costly (see pricing below) and requires a lengthy 3-4-day preparation and quality control process, with at least four major potential failure points. In contrast, the Plasmid2MC method utilizes conventionally and routinely produced source plasmids and operates entirely within a cell-free, in vitro system. This setup allows for precise timing and eliminates common failure points. As a result, Plasmid2MC offers a faster, more cost-effective, and more reliable alternative, consistently yielding higher-quality minicircle DNA compared to the MC-Easy™ kit. Notably, the intrinsic recombination reaction efficiency approaches 100%, with DNA losses primarily occurring during these purification steps, resulting in an overall protocol efficiency ranging from 56% to 34% for plasmid sizes ranging from 5.8 to 13.9 kbp, respectively. Using a 2.7 kbp bacterial backbone, the pure mcDNA conversion efficiency is an impressive 27% to 30% after accounting for the backbone size. Applying the Plasmid2MC method to produce a short 1211 bp mcDNA revealed that the recombination efficiency was maintained at 50.1%. However, due to the smaller ratio of mcDNA to the entire parental plasmid, the yield of mcDNA was reduced to 15.7%, which might still be practical in certain scenarios. For even shorter circular DNA constructs, the CV method, discussed further below, could prove more efficient. Although a conversion yield of 34–56% might initially seem low, it’s important to consider that we use conventionally produced, inexpensive, and plentiful plasmids as starting material. The advantage here is the straightforward and efficient production of high-purity mcDNA in quantities suitable for genome editing experiments. Our current pricing for a single custom project, which yielded 18.96 mg of ΦC31 integrase, was quoted at $1,552.95 by GenScript. At such a price, the custom order is less expensive than a 10-prep MC-Easy™ kit, which costs $1914 (Cat# MN925A-1) from System Biosciences Inc. With a recombination yield of 30%, this quantity of ΦC31 integrase would be sufficient to produce at least 5.6 mg of mcDNA, and potentially up to double that amount by using a higher excess of source plasmid to conserve ΦC31 integrase, making ordering a large quantity of ΦC31 cost-effective. Currently, ΦC31 protein is not available off-the-shelf in small quantities; however, one can anticipate its availability materializing with demand, akin to many commercially available proteins and restriction enzymes. Hypothetically, a kit containing 1% of this amount of ΦC31 integrase would cost $15.53 and could produce between 56 and 112 μg of mcDNA. Assuming that the production of ΦC31 integrase at industrial volumes would cost significantly less than what we paid, and considering that the recombination buffer contains no costly or hard-to-handle components, such a kit could be priced only slightly above this estimate. Even if the price of this hypothetical ΦC31 integrase kit were to quadruple from this estimate, it would still be competitively priced compared to many commercially available restriction enzyme and protein kits. Until such time, contracting a larger quantity of ΦC31 from a manufacturer like GenScript is an easy and cost-effective option, with the only drawback a longer delivery time. There are fundamental differences between the Plasmid2MC method and our previously published CV method38, which is primarily aimed at efficient fully synthetic substitution for plasmid-based guide RNA expression in CRISPR, base, and prime gene editing in cell cultures. The CV method offers rapid cell-free preparation of short circular DNA ready for transfection within 3 hours after receiving dsDNA from a commercial supplier, with minimal hands-on time38. In Oliynyk & Church 202247, we noted that error rates in commercially delivered dsDNA fragments were ~ 1/6000 per nucleotide for various commercial products48,49. This error rate is excellent for short CRISPR-Cas9 guide RNA sequences but may lead to a significant proportion of defective proteins when extending the circular DNA length to thousands of base pairs47 in the case of mcDNA. Additionally, the CV method performance is optimal at a CV size of 450 bp, where it reaches up to 62% efficiency in converting input linear double-stranded DNA into the circular expression vectors, diminishing to 30% efficiency at a 950 bp CV size and further to 5–8% efficiency at 1800 bp. This led us to define 450–950 bp as the practical range of the CV protocol47. The principal reason for this is that at longer linear source DNA sizes at practical concentrations, the CV ligation reaction favors polymerization into infinite chains of concatemers, with a vanishingly low proportion of circular DNA units50,51. The Plasmid2MC method addresses the aforementioned limitations of the CV method by leveraging E. coli amplification of a single plasmid clone, virtually eliminating DNA replication errors47,52,53. Our sequencing of recombined and purified mcDNA invariably showed a perfect match of the expected and achieved mcDNA sequences. Consequently, mcDNA produced by this method proves applicable not only for protein-expressing genome editing constructs but also for HITI templates and other large circular DNA constructs. Both methods can be used complementarily, as demonstrated in this study where CVs are utilized for guide gRNA expression alongside Cas9 and other proteins expressed by mcDNA in base editing validations. The Plasmid2MC method enables the production of larger quantities of mcDNA, which can be reused across multiple experiments, while the CV protocol efficiently generates multiple guide RNAs as needed. Our lab continues to explore genome editing by combining these two approaches. In most scenarios, kits like the QIAGEN Midiprep Plus ensure the removal of proteins and RNA during source plasmid preparation, whereas the Plasmid2MC protocol not only lowers endotoxin levels but also fully digests all unintended DNA. The FDA sets stringent requirements for plasmids and mcDNA intended for therapeutic applications in vaccines, demanding minimal bacterial genome and other DNA contamination, undetectable levels of protein and RNA, and endotoxin levels ≤0.04 EU/μg54. While FDA-certified mcDNA production would require well-documented equipment, procedures, and testing of each production lot by a medical or pharmaceutical provider, which was outside the scope of our research, the Plasmid2MC method can be used as a starting point for establishing such production. The Plasmid2MC method is capable of producing mcDNA that closely approaches these requirements, evidenced by a 35-fold reduction in bacterial endotoxins when using conventional plasmids produced with the help of QIAGEN Midiprep Plus as starting material. In contrast, we found orders of magnitude higher endotoxin levels in the mcDNA produced using the System Biosciences Inc MC-Easy™ Minicircle DNA Production Kit. In conclusion, the method’s ease of preparation, efficiency, and mcDNA purity make Plasmid2MC a valuable new tool for CRISPR-Cas9 and other applications that require bacterial backbone-free circular DNA. Materials and methods Plasmid2MC protocol The flow of the Plasmid2MC protocol is presented in Fig. 6. Plasmid2MC protocol flow. Preliminary steps Most of the reaction enzymes and buffers were based on commercially available kits and commonly used equipment. The exceptions include obtaining ΦC31 integrase and preparing the 2X reaction buffer as described below, and a small modification to the design of the plasmids. Obtain ΦC31 integrase protein from a reliable vendor. We subcontracted GenScript and received 18.96mg of ΦC31 protein at a concentration of 4.74 mg/ml (see Reagents and Materials). Obtain ΦC31 integrase protein from a reliable vendor. We subcontracted GenScript and received 18.96mg of ΦC31 protein at a concentration of 4.74 mg/ml (see Reagents and Materials). Dilute the ΦC31 integrase with pure glycerol to make a working buffer. Add glycerol to achieve a 50% vol/vol concentration suitable for storage at −20 °C in liquid form, sufficient for a few months of experiments and store the remaining ΦC31 integrase at −80 °C for long term storage. In our case, the ΦC31 integrase concentration in the working buffer became 2.6 μg/μl. Dilute the ΦC31 integrase with pure glycerol to make a working buffer. Add glycerol to achieve a 50% vol/vol concentration suitable for storage at −20 °C in liquid form, sufficient for a few months of experiments and store the remaining ΦC31 integrase at −80 °C for long term storage. In our case, the ΦC31 integrase concentration in the working buffer became 2.6 μg/μl. Prepare the 2X recombination buffer to contain: 200 mM NaCl, 40 mM Tris pH, 0.2 mM EDTA pH 7.5. For example, 50 ml of 2X buffer can be prepared as follows: a) 10 ml of 10 mM-Tris, 1 mM-EDTA (pH 8.0; Fisher Scientific Cat#BP2473) and a negligible addition of Tris to produce 2X of 0.1 mM EDTA.b) 2 ml of 1 M Tris (pH 8.0; Corning Cat#46-031-CM).c) 584 mg of NaCl.d) 38 ml of nuclease-free H2O (Invitrogen 10977015).Note: The literature lists a variety of buffer compositions and additives, such as TPP, BSA, and spermidine22,31,55. Our testing demonstrated that customarily used ingredients other than those listed above resulted in no effect—or even lower recombination yield—and appeared to be unnecessary. As demonstrated in the Results, we achieved complete recombination using our 2X buffer in overnight reactions, while ΦC31 protein was sufficiently stable to provide yield improvements over a 8-hour reaction versus a 4-hour reaction. Buffers with NaCl concentration in the range of 100–150 mM worked equally well, and the small increase of NaCl from the working buffer remained well within this range, even at the highest concentration of ΦC31 integrase (the Φ31 integrase storage buffer delivered by GenScript contained a relatively high NaCl concentration, see Reagents and materials). Prepare the 2X recombination buffer to contain: 200 mM NaCl, 40 mM Tris pH, 0.2 mM EDTA pH 7.5. For example, 50 ml of 2X buffer can be prepared as follows: a) 10 ml of 10 mM-Tris, 1 mM-EDTA (pH 8.0; Fisher Scientific Cat#BP2473) and a negligible addition of Tris to produce 2X of 0.1 mM EDTA. b) 2 ml of 1 M Tris (pH 8.0; Corning Cat#46-031-CM). c) 584 mg of NaCl. d) 38 ml of nuclease-free H2O (Invitrogen 10977015). Note: The literature lists a variety of buffer compositions and additives, such as TPP, BSA, and spermidine22,31,55. Our testing demonstrated that customarily used ingredients other than those listed above resulted in no effect—or even lower recombination yield—and appeared to be unnecessary. As demonstrated in the Results, we achieved complete recombination using our 2X buffer in overnight reactions, while ΦC31 protein was sufficiently stable to provide yield improvements over a 8-hour reaction versus a 4-hour reaction. Buffers with NaCl concentration in the range of 100–150 mM worked equally well, and the small increase of NaCl from the working buffer remained well within this range, even at the highest concentration of ΦC31 integrase (the Φ31 integrase storage buffer delivered by GenScript contained a relatively high NaCl concentration, see Reagents and materials). Make alterations to the design of source plasmids to includeΦC31 attP and attB sequences and add or identify unique restriction enzyme sites in the bacterial backbone area. The sequences are attB 5’-GTGCCAGGGCGTGCCCTTGGGCTCCCCGGGCGCG and attP 5’-GCCCCAACTGGGGTAACCTTTGAGTTCTCTCAGTTGGGGG, with the desired for circularization DNA sequence in between. Position them in a manner that allows the separation of the desired mcDNA and bacterial backbone elements into two circular units. The recombination occurs near the middle of the above sequences, with circularized DNA containing 5’-GCCCCAACTGGGGTAACCTTTGGGCTCCCCGGGCGCG, which may need to be considered when designing the mcDNA for a specific purpose.The bacterial backbone must contain a unique restriction enzyme site, which will be used to cut and digest the bacterial backbone while leaving the desired mcDNA intact. It is preferred to use an efficient restriction enzyme. As only three examples out of many possibilities, we prefer to use New England Biolabs (NEB) SspI-HF, NotI-HF, or BsaI-HFv2 since these are highly efficient and inexpensive enzymes. In this study, the bacterial backbone contained two SspI restriction sites, and we report the results using SspI-HF.Clone the plasmids as usual56–58, amplify using a competent E. coli, and extract DNA using a suitable kit for the required quantity of plasmid DNA. You can use our plasmid (deposited at Addgene as p-att-ef1a-ABE8e-dCas9-BSD #226116) for a bacterial backbone containing convenient restriction enzyme sites around attP and attB, as well as double SspI restriction enzyme sites and a unique NotI restriction enzyme site within the backbone. The sequence is also available in the Supplementary Data (p-att-ef1a-ABE8e-dCas9-BSD.gbk). Make alterations to the design of source plasmids to includeΦC31 attP and attB sequences and add or identify unique restriction enzyme sites in the bacterial backbone area. The sequences are attB 5’-GTGCCAGGGCGTGCCCTTGGGCTCCCCGGGCGCG and attP 5’-GCCCCAACTGGGGTAACCTTTGAGTTCTCTCAGTTGGGGG, with the desired for circularization DNA sequence in between. Position them in a manner that allows the separation of the desired mcDNA and bacterial backbone elements into two circular units. The recombination occurs near the middle of the above sequences, with circularized DNA containing 5’-GCCCCAACTGGGGTAACCTTTGGGCTCCCCGGGCGCG, which may need to be considered when designing the mcDNA for a specific purpose. The bacterial backbone must contain a unique restriction enzyme site, which will be used to cut and digest the bacterial backbone while leaving the desired mcDNA intact. It is preferred to use an efficient restriction enzyme. As only three examples out of many possibilities, we prefer to use New England Biolabs (NEB) SspI-HF, NotI-HF, or BsaI-HFv2 since these are highly efficient and inexpensive enzymes. In this study, the bacterial backbone contained two SspI restriction sites, and we report the results using SspI-HF. Clone the plasmids as usual56–58, amplify using a competent E. coli, and extract DNA using a suitable kit for the required quantity of plasmid DNA. You can use our plasmid (deposited at Addgene as p-att-ef1a-ABE8e-dCas9-BSD #226116) for a bacterial backbone containing convenient restriction enzyme sites around attP and attB, as well as double SspI restriction enzyme sites and a unique NotI restriction enzyme site within the backbone. The sequence is also available in the Supplementary Data (p-att-ef1a-ABE8e-dCas9-BSD.gbk). Reagents and materials A detailed inventory of the reagents and resources used in this research can be found in Supplementary Table S1. ΦC31 integrase protein was subcontracted from GenScript. We received 18.96 mg of protein, the currently quoted pricing for such order is $1,552.95. The product concentration was 4.74 mg/ml(by Bradford) A260/A280: 0.557, ≥85% purity, endotoxin level 0.338 EU/mg or equivalently 0.000338 EU/μg - effectively endotoxin free. See the protein amino acid sequence and QA in Supplementary Fig. S10. The GenScript ΦC31 integrase storage buffer composition was: 50 mM Tris-HCl, 500 mM NaCl, 10% Glycerol, pH 8.0. Protocol steps Recombine the source plasmid. Prepare the recombination reaction as listed in Table 2 below, with ΦC31 integrase and 2X buffer being half the reaction volume and plasmid DNA in water the other half. Recombine for at least 4 hours, preferably 8–12 hours or overnight. It is necessary to add proteinase K and digest ΦC31 integrase after the completion of recombination because both intact and degraded ΦC31 integrase negatively affect the column’s efficiency (consult the further section DNA purification after the completion of recombination). Recombine the source plasmid. Prepare the recombination reaction as listed in Table 2 below, with ΦC31 integrase and 2X buffer being half the reaction volume and plasmid DNA in water the other half. Recombine for at least 4 hours, preferably 8–12 hours or overnight. It is necessary to add proteinase K and digest ΦC31 integrase after the completion of recombination because both intact and degraded ΦC31 integrase negatively affect the column’s efficiency (consult the further section DNA purification after the completion of recombination). Use the spin column kit to clean up the recombined DNA. Use a spin column purification kit to perform intermediate purification of the recombination reaction output. We used QIAGEN’s QIAquick PCR Purification Kit59. Although the kit advertises use with up to 10 kbp DNA size, we found that it still performed well and resulted in cleaner DNA at 13.9 kbp when compared to other kits. Ensure you perform Proteinase K digestion on the ΦC31 integrase to prevent interference with DNA binding to the purification columns. For improved elution yield, heat the spin column according to the manufacturer’s recommendations. We warmed the purification columns in a heat block at 55 °C for 3 min before elution. For maximum elution efficiency, we used ≥60 μl of water. It is not necessary to measure the concentration of the eluted DNA at this step, unless desired. Typical elution yield is approximately 90% for short DNA fragments, but significantly less for larger plasmid sizes (see the note below). Use multiple columns if needed, based on the column binding capacity. Note: We validated the efficiency of the purification kits used for a similar purpose in Oliynyk & Church38,47, where we preferred QIAGEN’s QIAquick PCR Purification Kit (QIAquick)59. In this research, we additionally tested the yield of QIAquick and Takara’s NucleoSpin Gel and PCR Cleanup (NucleoSpin)60 with a 13.9 kbp plasmid. We found that with the input of 10 μg of the largest (13.9 kbp) plasmid used in this research, we retrieved 7.46 μg (74.6%) with NucleoSpin and 6.85 μg (68.5%) with QIAquick, using an elution volume of 55 μl in both cases. However, we found that when we used both kits for two-stage purification, the yield of NucleoSpin was lower, while the NucleoSpin DNA quality with preheating recommended by the vendor was consistently showing salt carry-over, with 260/230 quality typically in the 1.5–1.8 range, while the production runs of QIAquick exhibited excellent DNA quality. We achieved an even higher efficiency of > 85% large DNA retrieval rate using the QIAGEN QIAEX II Gel Extraction Kit; however, the DNA produced by this kit appeared to carry over resin particle pollution, and contamination indicated by an absorption peak at 225 nm with an amplitude exceeding the 260 nm DNA peak. QIAGEN does not recommend using DNA produced by QIAEX II kit for transfection, but recommends using the QIAquick kit instead, and we concur. Thus, we opted to use a QIAquick kit for a better combination of mcDNA final yield and quality. The two cleanups required have an estimated combined retrieval of the plasmid and mcDNA of 0.685 x 0.685 = 0.469 (46.9%)—i.e., 53.1% is lost in the column cleanup for this DNA size. This implies that with the maximum reported reaction yield of 33.9% for 13942 bp source plasmids—and assuming that recombination reaction efficiency is close to 100%—the additional pipetting losses and digestion of nicked and broken DNA constitute the remaining 13.0% difference. Use the spin column kit to clean up the recombined DNA. Use a spin column purification kit to perform intermediate purification of the recombination reaction output. We used QIAGEN’s QIAquick PCR Purification Kit59. Although the kit advertises use with up to 10 kbp DNA size, we found that it still performed well and resulted in cleaner DNA at 13.9 kbp when compared to other kits. Ensure you perform Proteinase K digestion on the ΦC31 integrase to prevent interference with DNA binding to the purification columns. For improved elution yield, heat the spin column according to the manufacturer’s recommendations. We warmed the purification columns in a heat block at 55 °C for 3 min before elution. For maximum elution efficiency, we used ≥60 μl of water. It is not necessary to measure the concentration of the eluted DNA at this step, unless desired. Typical elution yield is approximately 90% for short DNA fragments, but significantly less for larger plasmid sizes (see the note below). Use multiple columns if needed, based on the column binding capacity. Note: We validated the efficiency of the purification kits used for a similar purpose in Oliynyk & Church38,47, where we preferred QIAGEN’s QIAquick PCR Purification Kit (QIAquick)59. In this research, we additionally tested the yield of QIAquick and Takara’s NucleoSpin Gel and PCR Cleanup (NucleoSpin)60 with a 13.9 kbp plasmid. We found that with the input of 10 μg of the largest (13.9 kbp) plasmid used in this research, we retrieved 7.46 μg (74.6%) with NucleoSpin and 6.85 μg (68.5%) with QIAquick, using an elution volume of 55 μl in both cases. However, we found that when we used both kits for two-stage purification, the yield of NucleoSpin was lower, while the NucleoSpin DNA quality with preheating recommended by the vendor was consistently showing salt carry-over, with 260/230 quality typically in the 1.5–1.8 range, while the production runs of QIAquick exhibited excellent DNA quality. We achieved an even higher efficiency of > 85% large DNA retrieval rate using the QIAGEN QIAEX II Gel Extraction Kit; however, the DNA produced by this kit appeared to carry over resin particle pollution, and contamination indicated by an absorption peak at 225 nm with an amplitude exceeding the 260 nm DNA peak. QIAGEN does not recommend using DNA produced by QIAEX II kit for transfection, but recommends using the QIAquick kit instead, and we concur. Thus, we opted to use a QIAquick kit for a better combination of mcDNA final yield and quality. The two cleanups required have an estimated combined retrieval of the plasmid and mcDNA of 0.685 x 0.685 = 0.469 (46.9%)—i.e., 53.1% is lost in the column cleanup for this DNA size. This implies that with the maximum reported reaction yield of 33.9% for 13942 bp source plasmids—and assuming that recombination reaction efficiency is close to 100%—the additional pipetting losses and digestion of nicked and broken DNA constitute the remaining 13.0% difference. Cut the bacterial backbone DNA and digest it and all other non-circular or nicked DNA. Perform the reaction (as listed in Table 3) using a restriction enzyme unique to the bacterial backbone. Note: Perform a digestion test before performing this step for the first time to validate the time required for the complete digestion of byproducts with your DNA concentration and the chosen restriction enzyme cutting efficiency. The easiest method involves taking the quantity of the original plasmid equal to that used in a recombination reaction (much like in the example in this table), and incubating while taking samples for gel electrophoresis from the reaction volume every 20 minutes. When the gel shows complete digestion, purify the remaining reaction volume and verify on a gel that all of the DNA was digested. In our example using SspI or NotI restriction enzymes, the complete digestion was achieved within 40 minutes for 10 μg of plasmid DNA for all plasmid sizes we used in this study. The digestion takes longer—but not an excessive amount of time—for higher DNA concentrations. For example, for a 300 ng/μl digestion test reaction (30 μg in 100 μl reaction), almost complete digestion of a 13.9 kbp plasmid was achieved in 60 minutes, with no traces of DNA remaining at 90 minutes of digestion (see Supplementary Fig. S5). Cut the bacterial backbone DNA and digest it and all other non-circular or nicked DNA. Perform the reaction (as listed in Table 3) using a restriction enzyme unique to the bacterial backbone. Note: Perform a digestion test before performing this step for the first time to validate the time required for the complete digestion of byproducts with your DNA concentration and the chosen restriction enzyme cutting efficiency. The easiest method involves taking the quantity of the original plasmid equal to that used in a recombination reaction (much like in the example in this table), and incubating while taking samples for gel electrophoresis from the reaction volume every 20 minutes. When the gel shows complete digestion, purify the remaining reaction volume and verify on a gel that all of the DNA was digested. In our example using SspI or NotI restriction enzymes, the complete digestion was achieved within 40 minutes for 10 μg of plasmid DNA for all plasmid sizes we used in this study. The digestion takes longer—but not an excessive amount of time—for higher DNA concentrations. For example, for a 300 ng/μl digestion test reaction (30 μg in 100 μl reaction), almost complete digestion of a 13.9 kbp plasmid was achieved in 60 minutes, with no traces of DNA remaining at 90 minutes of digestion (see Supplementary Fig. S5). Use spin column kit to perform final clean up similar to Step 2 above. Note that the expected mcDNA yield is approximately 30% of the input DNA, and a lower column capacity will suffice. Adjust the number of QIAquick spin columns used accordingly. Use spin column kit to perform final clean up similar to Step 2 above. Note that the expected mcDNA yield is approximately 30% of the input DNA, and a lower column capacity will suffice. Adjust the number of QIAquick spin columns used accordingly. Table 2Recombination reaction example for 100 ng/μg (10.0 μg) of input of both plasmid DNA at 1.0 μg/μl and ΦC31 at 2.6 μg/μl in volume V = 100 μlReagentsComponentVolume (μl)Plasmid DNA, volume X to 10 μg10.0 (X)Nuclease-free H2O, to half-volume40.0 (V/2 - X)ΦC31, volume Y to 10 μg3.85 (Y)2X recombination buffer, to half-volume46.15 (V/2 - Y)Cycler scheduleStepTemperatureTimeRecombination30 °C8 hoursPause4 °CindefinitelyAdd 1.0 μl Proteinase K, digest ΦC3155 °C45–60 minutesPause4 °CindefinitelyRecommended concentration of DNA and ΦC31 integrase is 100–200ng/μl. For example, 200ng/μl of DNA and 100 ng/μl of ΦC31 (or 150 ng/ul of ΦC31 for smaller plasmids). Incubate at 30 °C for at least 4 hours, or preferably overnight. Notably, 30 °C is the optimal recombination temperature, and the reaction practically stops at 37 °C (see Fig. 2a). Be certain not to exceed 30 °C. Recombination reaction example for 100 ng/μg (10.0 μg) of input of both plasmid DNA at 1.0 μg/μl and ΦC31 at 2.6 μg/μl in volume V = 100 μl Recommended concentration of DNA and ΦC31 integrase is 100–200ng/μl. For example, 200ng/μl of DNA and 100 ng/μl of ΦC31 (or 150 ng/ul of ΦC31 for smaller plasmids). Incubate at 30 °C for at least 4 hours, or preferably overnight. Notably, 30 °C is the optimal recombination temperature, and the reaction practically stops at 37 °C (see Fig. 2a). Be certain not to exceed 30 °C. Table 3Example of the digestion/purification recombined DNA; reaction volume V = 100 μlReagentsComponentVolume (μl)DNA, volume X as eluted from Step 2(X)T5 Exonuclease (NEB M0663L)4.0Restriction Enzyme, e.g., SspI-HF (NEB R3132L)2.0rCutSmart 10x buffer (NEB B6004S)10.0Nuclease-free H2O (Invitrogen 10977015)as needed to 100.0 μlCycler scheduleStepTemperatureTimeDigestion37 °C60−90 minutesPause4 °Cindefinitely Example of the digestion/purification recombined DNA; reaction volume V = 100 μl For researchers using this protocol for the first time, we recommend using at least 40 μg of source plasmid for the initial experiment. In step 4 above, opt for a low elution volume of 20–25 μl. This will allow to become familiar with the protocol and, if necessary, adjust the production scale accordingly. The spin column purification steps described above are designed for the easy preparation of mcDNA in quantities ranging from 2 to about 30 μg. If larger quantities are required, scaling up the number of QIAquick columns may become impractical. In such cases, the cleanup steps can be performed more economically using isopropanol precipitation61 for step 2 of the protocol above. However, lower DNA extraction efficiency should be expected with methods other than modern spin columns. In a comparison study62, isopropanol precipitation reported DNA recovery of 52.47 ± 19.69% (see also63), centrifugal dialysis 12.58 ± 7.15%, and magnetic bead extraction 9.92 ± 3.89%, all notably lower than the above spin column efficiencies. Fine-tuning recombination buffer composition for optimal performance The mcDNA yield depends on the reaction buffer used, the source plasmid size, and a combination of DNA and ΦC31 integrase proportions and concentrations, recombination reaction temperatures, and recombination time. We explored this multivariate space while focusing on determining the optimal conditions for practical mcDNA yield. The work began with the validation of the buffers facilitating recombination. We tested the buffers used with ΦC31 integrase in other studies22,31,55, as well as a number of buffers customarily used for restriction enzyme reactions sourced from NEB. We found that all NEB buffers were not suitable for the recombination reactions, resulting in all DNA being completely digested in our preliminary tests of the protocol. The buffers containing only NaCl, Tris, and EDTA performed best, resulting in visually complete recombination of the DNA product in 12 hours (see Fig. 1d). We tested whether customarily used additives improve recombination efficiency. Formulations of the buffers with 1 to 5% glycerol had no noticeable difference in the recombination speed. The final recombination reaction mix ultimately contained 1% or more glycerol from the ΦC31 integrase working buffer. NaCl was most effective in the 100–150 mM range, with diminished performance at lower or higher concentrations. We also tested the addition of 100 μg/ml BSA and 5 mM spermidine—as used in22,55—and found that while complicating the handling, these additives had a slight negative effect on the recombination efficiency and thus were not used. Therefore, the final reaction concentrations were 20 mM Tris, 0.1 mM EDTA, 100 mM NaCl (used in the form of the 2X buffer described in the Methods), and quantities of ΦC31 integrase and plasmid DNA as required per experiment. This minimal buffer achieved complete recombination of the DNA product shown in Fig. 1d, thus it makes sense that additives beyond this would have no effect at best, and at worst interfere with the reaction. DNA purification after the completion of recombination Initially, we expected to develop a single-tube protocol where the recombination would be followed by the cutting of the bacterial backbone using restriction enzymes and digestion with T5 exonuclease within the same reaction tube, similar to what we successfully achieved for the ligation reaction of CVs (see Oliynyk & Church38). This would be particularly helpful because DNA purification spin column yields diminish with increasing DNA sizes. However, we found that the presence of ΦC31 integrase consistently reduced the on-target efficiency of the restriction enzymes we tested (NotI, SspI, and PsiI). Simultaneously, all of these enzymes exhibited star activity, leading to the deterioration of well-recombined mcDNA. We hypothesized that this behavior may be associated with ΦC31 integrase molecules binding with similar affinity to attP and attB, as well as to attL and attR32,33, while normally not effecting the reversal of recombination. It is possible that ΦC31 intermittently binds to other plasmid locations, or co-binds in the presence of restriction enzymes targeting specific DNA sequences and modifies their activity. Using the thermal inactivation of ΦC31 integrase at 80 °C has not abolished this behavior, and the use of thermolabile Proteinase K64 with subsequent heat inactivation has not completely resolved the issue, with a fraction of partially digested DNA fragments consistently present in Plasmidsaurus sequencing. After the above testing, it was determined that we could not dispense with the cleanup step after the completion of recombination. Additionally, the proteinase K digestion of ΦC31 integrase after recombination reaction was necessary before the column cleanup, which was also rendered less effective by ΦC31 integrase, with spin columns exhibiting clogging—particularly with higher ΦC31 integrase concentrations. As a result, the protocol included ΦC31 integrase digestion after the completion of recombination using proteinase K, followed by spin column cleanup of the DNA. This was followed by the cutting of the bacterial backbone with a suitable restriction enzyme and the digestion of all non-circular or nicked DNA using T5 exonuclease, followed by final spin column cleanup. DNA purification was performed using QIAGEN’s QIAquick PCR Purification Kit (QIAGEN Cat#28106) on a QIAGEN vacuum manifold and Eppendorf 5425 R microcentrifuge, per the manufacturer’s instructions. Storage conditions for mcDNA and ΦC31 Integrase The mcDNA intended for cell culture transfection is typically eluted in water. Recent studies indicate that freeze-thaw cycles have negligible effects on DNA integrity when stored in water or salt solutions65,66. However, to minimize potential risks, we avoid excessive freeze-thaw cycles, storing mcDNA at 4 °C for use within one week or at −20 °C, where DNA remains stable for many years (see Stable 16-year storage of DNA purified with the QIAamp DNA Blood Mini Kit67). The working buffer for ΦC31 integrase, stored in liquid form at −20 °C in 50% glycerol, was intended for ease of use and to avoid excessive freeze-thaw cycles68. We observed no differences in recombination efficiency within 1–2 months of storage. However, after more than 3 months, the recombination efficiency ΦC31 integrase in the 50% glycerol solution decreased by 10–15%. Thus, preparing aliquots of ΦC31 integrase in 50% glycerol, intended for use within 1–2 months, is preferred. Component concentrations in recombination experiments As shown in the Results, the mcDNA yield depends on the source plasmid sizes, as well as a combination of DNA and ΦC31 integrase proportions and concentrations, recombination reaction temperatures, and times. A thorough mapping of such multivariate space would take a large number of experiments. Our task was simplified by focusing on finding optimal practical concentrations useful for mcDNA production, which allowed us to rapidly pinpoint the useful reaction proportions. All experiments in Fig. 2 were performed using the Protocol steps described earlier, with the following particulars. The concentrations used were (numbers following components are in ng/μl): Fig. 2a, b: ΦC31 100, DNA 100; Fig. 2c, f: ΦC31 25–200, DNA 100; Fig. 2d, g: ΦC31 100, DNA 100–300; Fig. 2e, h: ΦC31 100, DNA 100; Fig. 2j: ΦC31 200, DNA 66.6. Figure 2i used all of the same data points displayed in Fig. 2c–h, presenting them as molar ratios of ΦC31 integrase molar concentrations relative to DNA molar concentrations. In this figure, ΦC31 integrase concentration was adjusted to account for 85% purity, as reported by GenScript quality analysis (Supplementary Fig. S10). The molar ratios were calculated based on ΦC31 integrase containing 611 residues and a corresponding molecular mass of 68 kDa. The molecular masses of corresponding plasmids were calculated based on plasmid size69, and the molar ratios were calculated as multiples of ΦC31 integrase molecules per DNA molecule for corresponding ΦC31 and DNA concentrations. CRISPR-Cas9 base editing: transfection, editing, handling, and sequencing of edited cells For all base editing validation experiments, we used plasmid p-att-ef1a-ABE8e-dCas9-BSD (see also Plasmid design and assembly further below) and mcDNA produced from this plasmid with our proposed protocol. For gRNA expression, we used the CV method described in Oliynyk & Church38. The CV sequences are available in the Supplementary Data. When needed, mcDNA was further vacuum-concentrated using Thermo Fisher Scientific SavantTM SpeedVacTM. Both the HEK293T and mESC lines were transfected using a Neon Transfection System (Thermo Fisher Scientific) using 10μl tips (Cat#MPK1096K). Cells were resuspended at 40–70% confluency, centrifuged at 300G, and resuspended in Neon Buffer R to accommodate 40k cells per edit—similar for both cell lines. At this point, the editing steps differed as follows: Editing and selection of HEK293T cells. We used 1400 ng of dCas9-ABE8e mcDNA (size 6833 bp) and dCas9-ABE8e plasmid (size: 9533 bp, endotoxins 0.57 EU/μg, Supplementary Fig. S19, S20), with 400 ng of the gRNA CVs to achieve a ~ 4X molar excess of guide expression vector to a protein expression vector, performing edits in triplicate, following38. The transfection was performed using a Neon protocol with two pulses at 1150 V and a duration of 20ms70. Edited HEK293T cells were seeded on 24-well plates (Corning) in DMEM supplemented with 10% FBS. Then, 16–18 hours after transfection, 10 ng/μl of blasticidin (Thermo Fisher Scientific) was used to select cells containing the base editor (based on a recommendation by Xiong et al.71). Selection continued for 2 days, noting that the control, non-transfected cells died 48 hours into the selection process; at this point the cells were passaged in media with 10 ng/μl of blasticidin. The media was changed with fresh media 24 hours later. Genomic DNA was extracted 120 hours after the beginning of selection using a Zymo Research Quick-DNA Microprep Kit (Cat#D3020). Genomic DNA processing and sequencing were performed similarly for both cell lines, as described further. Editing and selection of HEK293T cells. We used 1400 ng of dCas9-ABE8e mcDNA (size 6833 bp) and dCas9-ABE8e plasmid (size: 9533 bp, endotoxins 0.57 EU/μg, Supplementary Fig. S19, S20), with 400 ng of the gRNA CVs to achieve a ~ 4X molar excess of guide expression vector to a protein expression vector, performing edits in triplicate, following38. The transfection was performed using a Neon protocol with two pulses at 1150 V and a duration of 20ms70. Edited HEK293T cells were seeded on 24-well plates (Corning) in DMEM supplemented with 10% FBS. Then, 16–18 hours after transfection, 10 ng/μl of blasticidin (Thermo Fisher Scientific) was used to select cells containing the base editor (based on a recommendation by Xiong et al.71). Selection continued for 2 days, noting that the control, non-transfected cells died 48 hours into the selection process; at this point the cells were passaged in media with 10 ng/μl of blasticidin. The media was changed with fresh media 24 hours later. Genomic DNA was extracted 120 hours after the beginning of selection using a Zymo Research Quick-DNA Microprep Kit (Cat#D3020). Genomic DNA processing and sequencing were performed similarly for both cell lines, as described further. Editing and selection of mESCs. The initial transfection was performed with the same amount of DNA as above for HEK cells, with more cell mortality than expected. Thus, the DNA concentration was decreased by approximately four-fold, using 380 ng of dCas9-ABE8e mcDNA or 380 ng of plasmid with 100 ng of gRNA CV. This concentration resulted in the typical cell survival. The transfection was performed using a Neon protocol with two pulses at 1200 V and a duration of 20 ms72. Edited mESCs were seeded on 24-well plates (Corning) treated with Matrigel (Corning) in ESGRO-2i Medium (Millipore). For mESCs, we tested editing in quintuplicate with mcDNA and in duplicate with plasmid, as presented in the Results. The blasticidin selection was performed 14–16 hours after transfection with dose of 5 ng/μl, and same dose of 5 ng/μl was added 24 hours later and the cell cultures were passaged 48 hours after beginning of selection with fresh media. Genomic DNA was collected two days later in the same manner as for HEK293T cells. Editing and selection of mESCs. The initial transfection was performed with the same amount of DNA as above for HEK cells, with more cell mortality than expected. Thus, the DNA concentration was decreased by approximately four-fold, using 380 ng of dCas9-ABE8e mcDNA or 380 ng of plasmid with 100 ng of gRNA CV. This concentration resulted in the typical cell survival. The transfection was performed using a Neon protocol with two pulses at 1200 V and a duration of 20 ms72. Edited mESCs were seeded on 24-well plates (Corning) treated with Matrigel (Corning) in ESGRO-2i Medium (Millipore). For mESCs, we tested editing in quintuplicate with mcDNA and in duplicate with plasmid, as presented in the Results. The blasticidin selection was performed 14–16 hours after transfection with dose of 5 ng/μl, and same dose of 5 ng/μl was added 24 hours later and the cell cultures were passaged 48 hours after beginning of selection with fresh media. Genomic DNA was collected two days later in the same manner as for HEK293T cells. Genomic DNA processing and sequencing. Genomic areas of interest were amplified from 6 ng of whole genome DNA using HiFi Hot Start Readymix (Kapa) at 30 cycles according to the master mix protocol (see Supplementary Table S2 for the list of primers). This was followed by DNA separation via electrophoresis on 1% E-Gel EX Agarose Gels (Invitrogen) and extraction using a Monarch DNA Gel Extraction Kit (T1020S) with an additional washing step. Sanger sequencing was performed in duplicate for each physical sample using Azenta Genewiz with our PCR primers (Supplementary Table S2). The editing efficiency analysis was performed using EditR73,74 is summarized in Fig. 4 and illustrated in Supplementary Figs. S6, S7. Genomic DNA processing and sequencing. Genomic areas of interest were amplified from 6 ng of whole genome DNA using HiFi Hot Start Readymix (Kapa) at 30 cycles according to the master mix protocol (see Supplementary Table S2 for the list of primers). This was followed by DNA separation via electrophoresis on 1% E-Gel EX Agarose Gels (Invitrogen) and extraction using a Monarch DNA Gel Extraction Kit (T1020S) with an additional washing step. Sanger sequencing was performed in duplicate for each physical sample using Azenta Genewiz with our PCR primers (Supplementary Table S2). The editing efficiency analysis was performed using EditR73,74 is summarized in Fig. 4 and illustrated in Supplementary Figs. S6, S7. HITI editing: transfection, editing, handling, and sequencing of edited cells Transfection for HITI experiments. HEK293T cells were seeded into 24-well plates at a density of 120K per well 20–24 hours before transfection. At approximately 70% confluency, cells were transfected using Lipofectamine 3000 (Thermo Fisher Scientific) following the manufacturer’s instructions. Briefly, 250 ng of Cas9 plasmid (pCAG-1BPNLS-Cas9-1BPNLS, Addgene #87108), 200 ng of gRNA CVs, and 300 ng of HITI mcDNA template were mixed in 50 μl of Opti-MEM with 1 μl of 3000 Reagent and 1.5 μl of Lipofectamine 3000 Reagent per well. 48 hours later, cells were harvested by removing the media, washed one time with PBS, and then trypsinized using TrypLE Express (Thermo Fisher Scientific, Cat#12604013) for 3 minutes. Resuspended cells were pelleted via centrifugation for 5 minutes at 500 g. After centrifugation, cells were resuspended in media containing hygromycin (Thermo Fisher Scientific, Cat#10687010) at a concentration of 50 μg/ml and seeded in 6-well plates. Cells were allowed to grow for 3 days, then the hygromycin concentration was increased to 200 μg/ml. Continuous hygromycin treatment was carried out for 14 days until all unedited cells were eradicated, as measured by observing a control well with untransfected cells and treatment with the same hygromycin concentration. Transfection for HITI experiments. HEK293T cells were seeded into 24-well plates at a density of 120K per well 20–24 hours before transfection. At approximately 70% confluency, cells were transfected using Lipofectamine 3000 (Thermo Fisher Scientific) following the manufacturer’s instructions. Briefly, 250 ng of Cas9 plasmid (pCAG-1BPNLS-Cas9-1BPNLS, Addgene #87108), 200 ng of gRNA CVs, and 300 ng of HITI mcDNA template were mixed in 50 μl of Opti-MEM with 1 μl of 3000 Reagent and 1.5 μl of Lipofectamine 3000 Reagent per well. 48 hours later, cells were harvested by removing the media, washed one time with PBS, and then trypsinized using TrypLE Express (Thermo Fisher Scientific, Cat#12604013) for 3 minutes. Resuspended cells were pelleted via centrifugation for 5 minutes at 500 g. After centrifugation, cells were resuspended in media containing hygromycin (Thermo Fisher Scientific, Cat#10687010) at a concentration of 50 μg/ml and seeded in 6-well plates. Cells were allowed to grow for 3 days, then the hygromycin concentration was increased to 200 μg/ml. Continuous hygromycin treatment was carried out for 14 days until all unedited cells were eradicated, as measured by observing a control well with untransfected cells and treatment with the same hygromycin concentration. Flow cytometry analysis. Hygromycin-selected cells and untransfected control cells were harvested as described above. After centrifugation, cell pellets were resuspended into PBS with 2% FBS and 1 μg/ml of 4’,6-diamidino-2-phenylindol (DAPI) (Thermo Fisher Scientific, Cat#D1306) and transferred into 5 ml round bottom polystyrene flow tubes with cell strainer (Corning Life Sciences, Cat#352235). Cells were analyzed by flow cytometry (Symphony A1, BD Biosciences) and data was processed with FlowJoTM 10 software. An overview of gating and analysis is given in Supplementary Fig. S8. Flow cytometry analysis. Hygromycin-selected cells and untransfected control cells were harvested as described above. After centrifugation, cell pellets were resuspended into PBS with 2% FBS and 1 μg/ml of 4’,6-diamidino-2-phenylindol (DAPI) (Thermo Fisher Scientific, Cat#D1306) and transferred into 5 ml round bottom polystyrene flow tubes with cell strainer (Corning Life Sciences, Cat#352235). Cells were analyzed by flow cytometry (Symphony A1, BD Biosciences) and data was processed with FlowJoTM 10 software. An overview of gating and analysis is given in Supplementary Fig. S8. Genomic DNA extraction and PCR analysis. To confirm the correct integration of HITI templates, genomic DNA was extracted from HITI-treated cells and untransfected cells (as negative controls) using a Zymo Research Quick-DNA Microprep Kit (Cat#D3020). Genomic DNA processing and sequencing were performed as described above with slight modifications. ACTB or GAPDH-specific forward primers were used with a common mKate reverse primer to amplify the genomic DNA region spanning the junction of the HITI-mediated integration. Q5 Hot Start High-Fidelity 2X Master Mix (NEB., Cat#M0494) was used to amplify the genomic region using 30 cycles of amplification with 20 ng of total genomic DNA per reaction. PCR amplicons were cleaned up using QIAGEN’s QIAquick PCR Purification Kit (QIAquick) and separated by electrophoresis on 1% E-Gel EX Agarose Gels (Invitrogen). For sequencing, the Monarch DNA Gel Extraction Kit (T1020S) was used to purify the excised DNA band from the gel and sent for Sanger sequencing. Sanger sequencing was performed using Azenta Genewiz with a primer specific to mKate insert. All primers are listed in Supplementary Table S2. Genomic DNA extraction and PCR analysis. To confirm the correct integration of HITI templates, genomic DNA was extracted from HITI-treated cells and untransfected cells (as negative controls) using a Zymo Research Quick-DNA Microprep Kit (Cat#D3020). Genomic DNA processing and sequencing were performed as described above with slight modifications. ACTB or GAPDH-specific forward primers were used with a common mKate reverse primer to amplify the genomic DNA region spanning the junction of the HITI-mediated integration. Q5 Hot Start High-Fidelity 2X Master Mix (NEB., Cat#M0494) was used to amplify the genomic region using 30 cycles of amplification with 20 ng of total genomic DNA per reaction. PCR amplicons were cleaned up using QIAGEN’s QIAquick PCR Purification Kit (QIAquick) and separated by electrophoresis on 1% E-Gel EX Agarose Gels (Invitrogen). For sequencing, the Monarch DNA Gel Extraction Kit (T1020S) was used to purify the excised DNA band from the gel and sent for Sanger sequencing. Sanger sequencing was performed using Azenta Genewiz with a primer specific to mKate insert. All primers are listed in Supplementary Table S2. Plasmid design and assembly We used p-dCas9-SSAP-MS2-BB (Addgene Cat#183826)75 as a convenient source of bacterial backbone and dCas9 for Gibson assembly. The insert of the BSD gene is based on our earlier publication47, while the 10482 bp plasmid/mcDNA (named p-att-ef1a-MS2-RecT-dCas9-BSD, available at Addgene #226108) is used in our other ongoing project, and our dCas9-ABE8e plasmid (named p-att-ef1a-ABE8e-dCas9-BSD) is a modification of this plasmid with TADa for ABE base editing derived from11 and cloned using Gibson assembly NEBuilder HiFi DNA Assembly Master Mix (NEB Cat#E2621) and amplified using NEB 5-alpha F`Iq Competent E. coli (NEB Cat#C2992H). The necessary dsDNA fragments were ordered from Twist Bioscience. The plasmid p-att-ef1a-MS2-RecT-dCas9-BSD, 10482 bp in length, is a functional plasmid that was used as a basis for the recombination experiments (available at Addgene #226116). Two modifications of this plasmid—a truncation to create a 5758 bp plasmid p-att-ef1a-MS2-RecT-dCas9-BSD-truncated and an extension to create a 13942 bp plasmid p-att-ef1a-MS2-RecT-dCas9-BSD-extended—were used for recombination experiments but are not intended to be functional for genome editing. An additional even shorter 3871 bp parental plasmid p-att-SuperShort1211bp, designed to test the yield of 1211 bp mcDNA was produced from p-att-ef1a-MS2-RecT-dCas9-BSD-truncated plasmid by cutting it at AgeI and BbsI restriction sites and inserting a short DNA sequence to produce desired mcDNA length; similar to p-att-ef1a-MS2-RecT-dCas9-BSD-truncated, this plasmid is also designed exclusively for recombination testing purposes, and does not contain a valid protein expression vectors. The sequences for all aforementioned plasmids are available in the Plasmids sequences folder in the Supplementary Data. Testing of endotoxin levels To test the endotoxin levels for the source plasmid and the resulting mcDNA we used Charles River Endosafe® nexgen-MCS endotoxin testing equipment and PTS2001F Limulus Amebocyte Lysate (LAL) endotoxin testing cartridges certified for the FDA validation of endotoxin levels, helped by the expertise of the Wyss Institute at Harvard University that routinely performs such tests for biological materials intended for FDA validation. Per recommendation of Charles River technical support we used 10% dilution of Charles River BD100 dispersant, which is critical for reduction of the clumping and sticking of endotoxins in DNA samples, instrumental in achieving consistent and reliable testing results. The PTS cartridges contain 4 lanes each for every individual sample, two for spike calibration and two for the endotoxin level measurement and cross-validation, resulting in highly accurate measurements. For initial tuning, we used an earlier prepared with the use of QIAGEN Midiprep Plus p-att-ef1a-ABE8edCas9-BSD plasmid and resulting mcDNA produced with the help of Plasmid2MC. We found the source plasmid endotoxin level at 1.88 EU/μg and the Plasmid2MC produced mcDNA with 0.029EU/μg. This presented a 60-fold improvement in endotoxing levels (see the reports Supplementary Fig. S15). For this test we used the FDA approved cartridgec Cat#PTS2005F with the endotoxin range of 5–0.05 EU/ml, using dilution factor 2:1. This dilution allowed us to achieve an accurate reading for the mcDNA, however it showed that with such low endotoxin levels, the aforementioned cartridges PTS2001F with 1–0.01 EU/ml range were more appropriate, and we used them in all subsequent tests. The DNA was prepared in stock 1:1 concentrations of 1 μg DNA per ml, and the further necessary dilutions were performed as required per Charles River instructions. Endotoxin level tests were performed in triplicate. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Statistics and reproducibility All protocol reactions were performed in independent reaction duplicates (n = 2) and triplicates (n = 3) as stated in figures, and yield percentage values were averaged. The editing efficiency analysis was performed from Sanger sequencing using EditR73,74 from independent samples in triplicates (n = 3) for most of the samples and quintuplicates (n = 5) for mESC editing with mcDNA, with two technical replicates where applicable. The P values for editing efficiency calculated by EditR are listed in the Supplementary Data. The editing efficiency values were averaged. All endotoxin levels measurements were performed in triplicate using Charles River Endosafe® nexgen-MCS™ system and FDA certified Endosafe® LAL Cartridges PTS2001F and the test reports are presented in Supplementary PDF. Supplementary information Supplemental material PDF Description of Additional Supplementary Materials Supplementary Data 1 Reporting Summary Acknowledgements A.M. was supported by the Ibn Rushd Postdoctoral Fellowship from King Abdullah University of Science and Technology. E.K. received funding through SRA with Colossal Biosciences. We thank Shanda Lightbown from the Wyss Institute at Harvard University for providing access to equipment and assistance with endotoxin testing. Author contributions R.O. designed this research and performed Plasmid2MC and base editing experiments; R.O. and A.M. collaborated on conceptual foundations; E.K. defined mouse genomic targets for mESC base editing; A.M. conceived and performed HITI experiments and analysis; R.O., A.M. and G.C. analyzed results and wrote the manuscript. Peer review Peer review information Communications Biology thanks Qian Ren and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editors: Huan Bao and Mengtan Xing. Data availability The raw data for the results presented in this study can be found in a single SupplementaryData.ZIP file, composed of the following items: 1. Recombination Yields Table.xlsx. 2. Base Editing Efficiency Sanger-EditR Results.xlsx. 3. Base Editing plasmid, mcDNA and gDNA CV sequences. 4. HITI plasmid, mcDNA and gDNA CV sequences. 5. Sequenced Plasmidsaurus recombined mcDNA. Plasmids: p-att-ef1a-MS2-RecT-dCas9-BSD is deposited at Addgene #226108, and p-att-ef1a-ABE8e-dCas9-BSD is deposited at Addgene #226116. Competing interests GMC disclosures: https://arep.med.harvard.edu/gmc/tech.html. All other authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Supplementary information The online version contains supplementary material available at 10.1038/s42003-025-09157-7. References Associated Data Supplementary Materials Supplemental material PDF Description of Additional Supplementary Materials Supplementary Data 1 Reporting Summary Data Availability Statement The raw data for the results presented in this study can be found in a single SupplementaryData.ZIP file, composed of the following items: 1. Recombination Yields Table.xlsx. 2. Base Editing Efficiency Sanger-EditR Results.xlsx. 3. Base Editing plasmid, mcDNA and gDNA CV sequences. 4. HITI plasmid, mcDNA and gDNA CV sequences. 5. Sequenced Plasmidsaurus recombined mcDNA. Plasmids: p-att-ef1a-MS2-RecT-dCas9-BSD is deposited at Addgene #226108, and p-att-ef1a-ABE8e-dCas9-BSD is deposited at Addgene #226116.