Precise kilobase-scale genomic insertions in mammalian cells using PASTE

Abstract

Programmable gene integration (PGI) technologies are an emerging modality with exciting applications in both basic research and therapeutic development. Programmable addition via site-specific targeting elements (PASTE) is a PGI approach for precise and efficient programmable integration of large DNA sequences into the genome. PASTE offers improved editing efficiency, purity and programmability compared to previous methods for long insertions into the mammalian genome. By combining the specificity and cargo size capabilities of site-specific integrases with the programmability of prime editing, PASTE can precisely insert cargoes of at least 36 kb with efficiencies of up to 60%. Here, we outline best practices for design, execution and analysis of PASTE experiments, with protocols for integration of EGFP at the human NOLC1 and ACTB genomic loci and for readout by next generation sequencing (NGS) and droplet digital PCR (ddPCR). We provide guidelines for designing and optimizing a custom PASTE experiment for integration of desired payloads at alternative genomic loci, as well as example applications for in-frame protein tagging and multiplexed insertions. To facilitate experimental setup, we include the necessary sequences and plasmids for the delivery of PASTE components to cells via plasmid transfection or in vitro transcribed RNA. Most experiments in this protocol can be performed in as little as 2 weeks, allowing for precise and versatile programmable gene insertion.

Introduction

The need for programmable gene integration (PGI) technologies for insertion of large nucleic acid cargos is driven both by potential therapeutic applications and uses in basic research. Despite significant progress in translating CRISPR-based genome editing therapies for clinical purposes¹, there are still significant barriers to applying these to many patients and experiments. Many diseases are highly genetically heterogeneous, with hundreds or even thousands of mutations sometimes associated with the same disease with similar phenotypic presentation. Cystic fibrosis, for example, has been associated with approximately 2000 mutations in CFTR². Insertion of large DNA sequences through PGI could replace or supplement pathogenic allele(s) by insertion of large portions of the healthy allele as cDNA, serving as a ‘one-size-fits-all’ therapy for common and highly heterogeneous diseases. Moreover, PGI technologies enable integration of complex responsive circuits³ for myriad cell therapy applications and offer unprecedented ability to study and manipulate the genome⁴.

Many common genome engineering approaches for inserting long DNA sequences leverage double-stranded breaks (DSB) and stimulation of cellular DNA repair mechanisms such as non-homologous end joining (NHEJ)⁵, alternative end-joining (Alt-EJ), or microhomology-mediated end-joining (MHEJ)^5–10. While these approaches are useful in many contexts, DSB-based approaches have multiple downsides, including small insertions/deletions (indels) at the target site, off-target insertions at unintended DSBs, activation of p53, chromosomal rearrangements, lack of control over orientation of insertion, and low efficiency and/or toxicity in some cell types such as neurons and stem cells^11–14.

Motivated by these issues, gene editing technologies have been developed to circumvent the requirement for DSB generation, including base and prime editing^15–17. Base editors (BEs) fuse adenine or cytidine deaminases with a nickase mutant Cas9 (nCas9), which cleaves only a single strand of DNA. Adenine- or cytosine BEs (ABE or CBE) install A-to-G (or a T to C change on the opposite strand) or C-to-T (or a G to A change on the opposite strand) transition mutations within a short editing window, respectively, dictated by an sgRNA. Additional BE variants install other edits, including C-to-G transversions and other changes dependent on specific repair processes^16–18. Prime editing (PE), combining a nickase, such as nCas9, with a RT domain, such as the moloney murine leukemia virus reverse transcriptase (M-MLV RT)¹⁵, further expands the scope of potential edits. PE innovates by providing arbitrary sequence edits as part of a modified guide RNA, called a prime editing guide RNA (pegRNA). The pegRNA contains the requisite spacer sequence to direct genomic nicking, allowing for priming from this genomic nick by the RT on the pegRNA template to generate DNA of the desired edit. Addition of this sequence results in a flap containing the edit, which is resolved by endogenous DNA repair mechanisms. Prime editing efficiencies can be improved through modulation of endogenous factors such as the mismatch repair pathway^19,20 or end protection of the exposed 3’ end of the pegRNAs^21,22. PE enables all possible base edits (all 4 base transitions and 8 transversions), small insertions (~1–100 bp), and deletions (~1–80 bp), by encoding these changes within the pegRNA. Further developments of PE, such as twinPE, use a pair of pegRNAs targeting the sense and antisense strand of DNA, leading to enhanced editing efficiency and the ability to install larger insertions and deletions (~200 bp)^23–26. However, these approaches still struggle with integrating kilobase-scale DNA sequences efficiently.

To address the outstanding need for programmable gene insertion and replacement independent of double-stranded break DNA repair pathways, our lab recently developed a PGI approach: Programmable Addition via Site-specific Targeting Elements (PASTE)²⁷. PASTE enables the efficient insertion of DNA sequences without known size restrictions (insertions of 36 kb have been achieved thus far) by combining prime editors with large serine recombinases, such as the integrase from the Bxb1 mycobacteriophage, either in fusion or in trans. PASTE eliminates the need for an exposed DSB and inserts diverse payloads into the genome with high fidelity and very low off target editing, with the capability for multiplexed concurrent insertions. PASTE is complementary to another method, TwinPE-knockin, which similarly uses prime editing and Bxb1 integrase to insert large DNA cargoes with high efficiency²³.

PASTE combines the programmability of PE with the insertion capabilities of serine integrases such as the Bxb1 integrase²⁸ through two molecular events. First, the prime editor utilizes a pegRNA containing the guide sequence, a 3′ extension reverse transcriptase template (RTT) encoding the desired edit, such as an attB target sequence of the integrase, and a primer binding site (PBS) which is complementary to the target genomic sequence and allows priming and reverse transcription of the insertion site into the genome¹⁵. We refer to pegRNAs encoding the insertion of attB attachment sites into the genome as attachment site-containing guide RNAs (atgRNAs). Within the cell, the nCas9 utilizes the spacer to scan the genome for the target site of insertion, where it binds and cleaves (‘nicks’) a single strand of DNA. The PBS hybridizes to the exposed DNA strand to form an RNA-DNA complex that can be used to prime the reverse transcription of the atgRNA 3′ extension by the M-MLV RT. This forms a DNA flap that contains the desired edit (i.e., the attB site), which can be incorporated by DNA repair mechanisms into the genome. Additionally, a nicking guide RNA targeting the opposite strand to the insertion can be used, which biases cellular DNA repair pathways to maintain the desired edit in the genome rather than editing it out. One can alternatively utilize twinPE to insert the attachment sites, where twin atgRNAs encoding the attachment sites are each targeted to the sense and antisense strand of the target site²³. Because of its use of PE, PASTE can leverage PE optimizations, including pegRNA optimizations^22,23, protein engineering²⁹, and DNA repair pathway modulation to improve activity^19,20.

In the second stage of PASTE, the integrase facilitates insertion of a payload into the target sequence. The serine integrase used in PASTE, typically Bxb1 integrase, is expressed either fused to the C terminal end of the PE machinery (PASTEv2/PASTEv3) or expressed separately via a self-cleaving 2a sequence (PASTEv1). While others have reported successful prime editing and TwinPE knockin with components delivered in trans^15,23, we find that the optimal construct is situation specific and as a general recommendation suggest screening of both systems when attempting to insert at a new target. Once expressed, integrase dimers bind to the genomic attB and payload attP sites to assemble a tetrameric synaptic complex³⁰. Catalytic serine residues then nick the DNA adjacent to the central dinucleotides of the attachment sites, generating two half sites with central 2-bp overhangs. Rotation of the complex subunits swap the half sites, which hybridize via their complementary central dinucleotides and are ligated together^28,31,32. This process inserts the DNA payload into the genome flanked by stereotyped attL and attR sites, genomic scars similar in size to the original attachment sites. Depending on the application, the genomic scars may be inconsequential, as payloads for gene or cell therapies can be inserted in non-coding or intronic regions where the scars are minimally disruptive. For example, insertion into the first intron of a gene with a payload containing a 5’ splice acceptor site and subsequent exons would not be affected by genomic scars. However, certain situations that require direct correction of coding sequence may not be suitable for PASTE, as genomic scars would be contained within the reading frame and potentially disrupt protein function. While PASTE inherently involves two molecular events, all components are delivered to cells in a single step.

In this protocol, we provide design considerations and practical guidelines for the user to perform a PASTE experiment in cultured cells, including the design, cloning and transfection of the different PASTE components in cultured cells, plus NGS and ddPCR readouts of integration, all of which can be carried out in the space of 2–3 weeks (Fig. 1).

Figure 1: Overview of workflow for implementing PASTE for a new application.

An outline of the experimental process for designing, generating, testing, and using PASTE at a novel locus or for a new application. Timing for each block of steps is shown at the left. The protocol can be loosely split into 4 stages, the first being design and generation of necessary constructs, either pegRNAs only or pegRNAs and nicking guides, cloning of custom payloads, and purification and sequence verification of all constructs. The second stage is to screen combinations of atgRNAs for attB site insertion by amplicon sequencing. The third section covers transfection to read out integration efficiency with EGFP or custom payloads. The final portions of the protocol cover how working constructs can be generated as viral vector or RNA template for applications of PASTE depending on the model cell type or use case.

Comparison to other methods for long insertions

PASTE joins a toolbox of technologies for inserting long DNA sequences (Fig. 2). TwinPE knock-in can effectively insert an attB followed by integration of large payloads by Bxb1 integrase²³, with subtle differences from PASTE: in TwinPE knock-in, prime editing and integrase plasmids are delivered as separate plasmids, in contrast to PASTE, where the editing machinery is fused via an XTEN linker or with a self-cleaving P2A sequence. Furthermore, TwinPE and PASTE differ in the length of the Bxb1 integrase attachment sites used. Additional dual pegRNA methods have been reported in the literature, such as PRIME-Del³³, PEDAR²⁶, HOPE³⁴, GRAND²⁴, Bi-PE³⁵ and TJ-PE²⁵. In TwinPE and GRAND, the sequences reverse transcribed from the dual pegRNAs are complementary and thus anneal, replacing the sequences in between the nick sites. PRIME-Del and Bi-PE likewise use template sequences that are complementary after reverse transcription and anneal. However, they also match the genomic sequence upstream of the nick on the opposite strand. HOPE differs from these methods in that the inter-nick sequences are not deleted. In addition to TwinPE like methods, other insertion approaches have harnessed type V-K CRISPR-associated transposases (CASTs) for large DNA insertions in bacterial cells^36,37, but with limited efficacy in eukaryotic cells^38,39.

Figure 2: Comparison of PASTE to other technologies for programmable DNA insertion.

Overview of steps and components involved in generating a large insertion with Cas9 and homology directed repair (HDR)/Non-Homologous End Joining (NHEJ), TwinPE/GRAND editing, PASTE, or CRISPR-associated transposases (CAST). With Cas9 HDR or NHEJ, a Cas9 effector is recruited to a genomic target using one or multiple gRNAs in order to create a double-strand break. The break is repaired through HDR with high fidelity using a supplied dsDNA donor targeted via homology, or via NHEJ for error-prone correction. TwinPE inserts sequences using a variation of prime editing, with a pair of pegRNAs encoding short sequences in their tails. In PASTE, a prime editing event installs an attB site using an atgRNA and nicking guide (paired atgRNAs if using twinPE.) The inserted attB site serves as a beacon for Bxb1 integrase to integrate the supplied attP-containing template sequence, which can be tens of thousands of base pairs long. Finally, CAST uses a gRNA and either a type V-K or I-F Cas effector to recruit a set of transposon proteins (tniQ, tnsB, tnsC in the case of V-K systems) which transpose a supplied dsDNA template approximately 60 bp downstream of the gRNA target.

Alternative site-specific recombinases than Bxb1 integrase for PASTE editing

A variety of enzymes can be used to catalyze the integration event of PASTE. While some commonly used proteins from elsewhere in molecular biology (such as members of the tyrosine recombinase family) lack the specificity or efficiency of integration necessary for repurposing in this manner, others can have desirable properties. For example, genome mining of different serine recombinases revealed a variety of enzymes with smaller attachment sites than the Bxb1 integrase^27,40. This can have a major impact for genomic regions where prime editing rates are low, because the installation of smaller sequences in the range of ~30–36 bp is more efficient than that of the larger Bxb1 integrase attB sites. Experimentally the minimal attB sequence for Bxb1 was determined to be 38bp⁴¹, but genomic mining efforts have identified other serine integrase attB sites as short as 26bp⁴⁰. Depending on the target, overall editing rates can be a balancing act between integrase efficiency (generally better with a larger attB sequence) and prime editing efficiency (generally better with smaller insert).

Screening of novel integrases has revealed that integrase efficiency can vary by locus, possibly due to the surrounding genomic context or chromatin structure. Metagenomic mining and screening of novel integrases has also identified integrases capable of naturally integrating into sites in the human genome⁴⁰. Despite this wealth of integrase sequences, reprogramming integrases to desired sites without introduction of a landing pad remains elusive. Testing several different integrases when optimizing a particular application of PASTE could potentially improve the overall editing efficiency.

Applications of PASTE editing

The primary applications of PASTE are the replacement or supplementation of genes in the human genome, for example as a potential therapy for a genetic disease. For gene supplementation, the user can insert a healthy gene allele, such as a cDNA with a 5′ splice acceptor site, into the native locus, for example in the first intron, to ensure splicing into the wild-type transcript. Similar approaches have been previously employed in cells using CRISPR-Cas9 knockin with incorporation of a repair template⁴², though this comes with numerous downsides associated with the DSBs that may render them clinically intractable (e.g., indel formation, chromosomal rearrangements, and activation of p53), as well as issues with repair efficiency in non-dividing cells or the bi-directionality of the insert. A variant of PASTE known as PASTE-replace can be employed if the goal is PGI along with a deletion of a region of the genome, allowing for true sequence replacement. PASTE-replace utilizes the same principle as PRIME-Del³³, with atgRNAs that flank the region to be deleted and contain overlapping sequences in the reverse transcription templates. Methods similar to PASTE have also been applied to non-human organisms, such as rice plants⁴³.

Limitations

The rate-limiting step of performing a successful PASTE experiment is often the identification of efficient atgRNAs for insertion of the attB at the desired locus. atgRNA efficiency is highly dependent on the choice of spacer sequence and length of PBS and RT sequences. While there are algorithms for predicting effective pegRNA sequences^44,45, we have found that there is no substitute for experimentally screening atgRNAs, which can be expensive and time consuming. We recommend searching for optimal atgRNAs using cell lines that are easy to culture and transfect with high efficiency, such as HEK293FT cells for human loci or N2As or Hepa-1–6 cells for mouse targets, as it is usually the case that an optimal atgRNA is generalizable to other cell types. One caveat is that delivery of atgRNAs as a DNA plasmid in their present design will result in integration of the payload into the attB sites encoded on the atgRNA plasmids. Despite this, we have routinely achieved high editing efficiencies when delivering the atgRNAs as plasmids, though the user can mitigate this by delivering them as RNAs, if desired.

An additional limitation of PASTE is that, due to the mechanism of payload insertion, small stereotyped attL and attR scars are left behind in the human genome (Fig. 3). While these can be leveraged as a linker sequence to tag an ORF in frame, these are exogenous sequences not naturally found in the human genome, which, although unlikely, could have unintended outcomes. Finally, PASTE utilizes three enzymes (Cas9, M-MLV RT and Bxb1 integrase) in addition to 1 or 2 atgRNAs and a payload plasmid, resulting in a large amount of material that must be delivered to cells. Single viral vector delivery is possible only with adenoviral vectors, due to cargo size constraints. If delivery with adeno-associated (AAV) vectors is desired (see Box 1), then a dual vector system is necessary, which can have a negative impact on editing efficiency. Additionally, the payload plasmid must be delivered as DNA to be inserted by Bxb1 integrase, meaning that payload delivery is challenging with lipid nanoparticles (LNPs). Delivery with retro viruses (e.g. lentivirus) is possible due to reverse transcription of the payload to cDNA, but may come with additional drawbacks such as integration into the genome; delivery with integration-deficient lentiviral vectors (IDLVs)⁴⁶ may be overcome this concern, although IDLVs have not yet been tested with PASTE.

Figure 3: Sequence level view of PASTE editing at an example locus.

Outline showing each of the steps involved in PASTE insertion of a payload into a genomic locus (NOLC1 shown) with sequence level resolution. The atgRNA used is shown linearized at the top to illustrate the relative positions and orientation of each of the elements. The first stage of the process involves an initial prime editing event where the bound atgRNA recruits a nickase-cas9 prime editing construct with fused M-MULV and Bxb1 integrase (gray, background) to a site containing the protospacer (blue) and linked directly adjacent PAM (light orange). The attB sequence (dark orange) is reverse-transcribed off of the hybridized PBS domain (gold). A separate spacer (green) and prime editing effector (not shown) creates a nick that allows for resolution of the “flap” in the direction of the edited product. The second stage of the process is catalyzed by the integrase domain of the PASTE editor and involves paired attB/attP sites (dark orange, cyan) with matched central dinucleotide sequences (dark purple). The Bxb1 integrase binds to the attB and attP sequences on the genomic target and payload respectively and catalyzes a set of cleavage and rotation events that ultimately connect the left side of the attB to the right side of the attP (as shown) to form the attL, and the left side of the attP and right side of the attB to form the attR. The result is an integrated payload flanked by upstream attL and downstream attR scar sequences. RT = reverse transcriptase; M-MULV = Moloney Murine Leukemia Virus; atgRNA = attachment site containing guide RNA; PBS = primer binding site; PAM = protospacer adjacent motif.

(BOX 1). Viral template production.

Two types of viral DNA vectors can be employed for template delivery: 1) AAV, and 2) adenovirus. The advantage of viral delivery for the DNA template is it transports the DNA template to the nucleus of the cell, which can be especially important in the case of non-dividing cells. In this box, we describe design and production strategies for both the AAV and adenoviral templates for PASTE, which can be used in place of the PASTE plasmid transfection in Steps 35–40.

AAV can be produced at small scales using triple transfection of the helper and template plasmids in HEK293T or HEK293FT cells. Using low passage cells is encouraged for efficient production. When designing viral templates, note that a promoter is not needed if inserting the template at the endogenous gene location with the intent to replace expression of the mutated gene copy. In this case, either design the viral template to be inserted at the end of the endogenous promoter region or in the first intron of the gene with a splice acceptor to allow for splicing onto the first exon.

AAV Plasmid transfection procedure:

1. Thaw a fresh aliquot of HEK293FTs (approximately 1×10⁶ cells) in warm DMEM with 10% FBS, 1% v/v Penicillin/Streptomycin and culture in a T25 flask or 10 cm dish.

2. 24 hours before transfection, expand HEK293FTs according to manufacturer protocols. For a small-scale production, 2–4 T225s are needed for transfection. AAV production efficiency can vary by serotype, but a relatively small-scale production like this should produce >1×10⁸ viral genomes/mL

3. For each AAV construct to be generated, transfect a total of 90 μg of total DNA per T225. Combine packaging and capsid plasmids with each transfer plasmid in a 1:1:1 molar ratio.

4. Combine with 3 mL of OptiMEM, vortex, and then add 270 μL of PEI at 1 μg/μL and vortex.

5. Wait for 15 minutes, then aspirate the media from the cells and replace with fresh media containing the transfection mix.

6. 72 hours after transfection, collect media and ultracentrifuge or concentrate through a 10 kDa MWCO filter. Resuspend using PBS. A more thorough purification that includes cell lysis (and therefore would dramatically increase yield) can also be performed according to Addgene’s suggested process (https://www.addgene.org/protocols/aav-purification-iodixanol-gradient-ultracentrifugation/)

7. Titer calculation: treat purified AAV with DNAse to break down un-encapsulated genomes or plasmid DNA by adding 0.5 μL of DNAse to 1 μL of virus, along with 2uL of DNAse buffer and 16.5 μL of H₂0 for a total of 20 μL. The DNAse digestion should be heated at 37°C for 1 hour, then 75°C for 15 minutes.

8. Proteinase treat the DNAse-digested virus to break open the capsids by adding 1 μL of Proteinase K to 5 μL of sample and 14 μL of H20 for a total of 20uL. Thermocycle the reaction for 30 minutes at 50°C, then 10 minutes at 98°C to denature the enzymes.

9. Set up a SYBR Green qPCR reaction using primers that bind within the transfer backbone and compare to a standard curve of successive 1/10 dilutions of the plasmid used to produce the AAV. Genome copies can be calculated based on the concentration and size of the plasmid backbone. For further information please see: https://www.addgene.org/protocols/aav-titration-qpcr-using-sybr-green-technology/.

Adenovirus plasmid transfection procedure:

When designing adenovirus templates, one can either use the popular adEasy^™ system⁶⁴ which only allows for ~8 kb of insert sequence or the helper-dependent adenovirus (AdV) system, which allows for maximal 36 kb sequences. Here, we briefly describe the adEasy production protocol due to the relative ease with which it can be performed.

1. Construction of AdV Plasmid (~1 week): Utilize the AdEasy^™ system⁶⁴, which involves two plasmids: a shuttle vector for cloning the transgene and pAdEasy^™ containing necessary adenoviral genes. Analyze the final purified plasmid stock with endonuclease restriction digest or full sequencing to confirm integrity.

2. Initial Production (2–3 weeks): Transfect HEK293 cells with the AdV plasmid construct and maintain in culture at 37°C and 5% CO₂ for up to 20 days⁶⁴.

3. When cells are fully confluent, harvest them, and release the virus by multiple freeze-thaw cycles to obtain primary low titer AdV stock. Store the resulting lysate at −80°C or use immediately for amplification. Do not change the media (add fresh media only once a week) and avoid harvesting cells before at least 10 days to ensure adequate virus titer.

4. Amplification (1–2 weeks): Infect more HEK293 cells with AdV stock to produce a higher titer virus. Harvest the virus once about 50% of the infected cells show cytopathic effect, indicated by rounded and clumped cells.

5. Repeat amplification 2–4 times, each round increasing the virus quantity by 10–100 fold.

6. Purification (2 days): Purify AdV for in vivo use, typically using density gradients and ultracentrifugation. Basic protocols for a simple purification scheme using ultracentrifugation with CsCl or ATPS can be found here⁶⁵, while a separate protocol for adenoviral production that focuses on CsCl purification can be found here⁶⁴. After purification, dialyze the high-titer AdV against PBS or other preferred storage buffers using a 100kDa MWCO concentrator (Amicon, Sigma-Aldrich) and store at −80°C.

7. Titer Calculation (1 or 10 days): Measure physical titer (AdV genomes/mL) using a quantitative PCR kit with viral DNA specific primers, such as the Takara Adeno-X qPCR Titration Kit. Use Ad5 reference material (ATCC#VR-1516) as an internal control for assay validation. Transfection at the scale of 2x 25cm dishes should produce in the range of 1×10^8–10 pfu/mL, while subsequent rounds of infection and purification should increase the functional virus to 1×10^12–13 pfu/mL⁶⁴

For LNP delivery of PASTE machinery, the protein components of PASTE are encoded as single or dual mRNAs and the guides are delivered as synthetic guide RNAs, allowing for encapsulation of all these RNA components in a single lipid nanoparticle vector. This LNP formulation must then be combined with a DNA vector for template delivery, such as adenovirus or AAV, for a combined LNP and template delivery approach.

Experimental design considerations

To design a successful PASTE experiment, we recommend systematically addressing key components, focusing on the following aspects, discussed further in the sections below: (i) choosing between attB site insertion methods, particularly TwinPE and PE3, while considering their relative efficiencies and drawbacks; (ii) selection of which PE system and PASTE architecture to achieve maximum integration efficiency in the desired experimental model; (iii) comprehensive design and optimization of atgRNAs, including pegRNA components, to enhance attB insertion efficiency; (iv) payload design, taking into account factors like insert size, frame, and the choice of inserting either the payload or the entire plasmid, as well as the incorporation of specific sequence features; (v) comparing and contrasting next-generation sequencing (NGS) and droplet digital PCR (ddPCR) as methodologies for assessing integration efficiency, with a focus on their strengths and limitations; and (vi) evaluating various delivery methods in terms of their efficacy and compatibility with PASTE components.

For simplicity, in this protocol we always place the attB in the genome and the attP in the payload. However, the user has the option of using the inverse arrangement (attP in the genome and attB in the payload), which may have higher efficiencies at ceratin loci and reduce offtarget insertions. Several PASTE versions have been developed: PASTEv1 has a P2A linker for trans Cas9-RT and Bxb1 integrase expression, PASTEv3 has a (GGS)6 linker for fused Cas9-RT and integrase expression, and PASTEv4 uses the B.cereus integrase in place of Bxb1 (ref. 27), Consideration of the version of PASTE is locus and cell type specific, as there are no current prediction models for the optimal version of PASTE in particular contexts. In addition, as PASTE is applied in different cell types and at different loci, specificity must be evaluated. While PASTE is specific in HEK293FT cells when evaluated with two genomic loci targets, more rigorous specificity analysis is needed with deeper sequencing across different cell types to ensure that integration or recombination off-targets are not occurring, even at rare frequencies.

Comparing prime editing methods for attachment site insertion

To perform PASTE experiments, the user can elect to insert attB sites by TwinPE²³ or other variants of prime editing such as PE2 or PE3 (ref. 15). In most applications, TwinPE is the preferrable method for attB insertion, as it is generally more efficient than PE3. Furthermore, screening combinations of atgRNAs is easier with TwinPE, as design considerations lie only in the spacer sequence and length of PBS; PE3, for example, adds an extra design consideration in the length of the RTT and that the nicking guides can be on either side of the atgRNAs. Furthermore, TwinPE atgRNAs are shorter and therefore cheaper to synthesize as RNA, if desired, enabling many applications that require RNA delivery.

atgRNA design

The first step of PASTE is to insert attB sites by prime editing, which requires the identification of efficient pegRNAs containing the attachment site (atgRNA). Generic pegRNAs have 3 different sequence considerations: the spacer, which programs the location of the nCas9 ssDNA cut site; the reverse transcription template (RTT), which contains both the edit to be introduced in the genome through reverse transcription and the homology to the target site, and the primer binding sequence (PBS) which primes the template reverse transcription (Fig. 3). For atgRNAs specifically, the PBS, RTT and spacer sequences vary and can all be optimized to reach maximal attB insertion efficiency. However, the specific insertion template within the RTT encoding the Bxb1 integrase attB, is generally not varied, except in some cases such as where one wishes to use alternative integrases or to multiplex integration via variation of the central dinucleotide (see Box 2).

(BOX 2). Multiplexing PASTE with central dinucleotide engineering.

The central dinucleotide is critically involved in recombination between attachment sites by Bxb1 integrase and can confer specificity between the attB and attP. Attachment sites containing certain central dinucleotides will only recombine if the corresponding attachment site has the same central dinucleotide, with minimal compatibility wbetween attachment sites containing different dinucleotides. This specificity facilitates multiplexed editing by leveraging attB and attP sites containing orthogonal central dinucleotides, resulting in simultaneous edits at different sites. We have found that the dinucleotides that exhibit the strongest self-selectivity with minimum cross-talk with other dinucleotides are GA, AG, AC and CT. In our original paper, we leveraged these specificities at three sites, using:

1. An ACTB targeting pegRNA inserting an attB with CT dinucleotide, with matching CT dinucleotide attP donor packaging GFP

2. An LMNB1 targeting pegRNA inserting an attB with AG dinucleotide, with matching AG dinucleotide attP donor packaging mCherry

3. A NOLC1 targeting pegRNA inserting an attB with GA dinucleotide, with matching GA dinucleotide attP donor packaging YFP

If more multiplexing is required beyond four orthogonal sites, we have profiled all possible dinucleotide combinations in our original manuscript and additional dinucleotides with minimal crosstalk can be identified and used²⁷.

For TwinPE, only the spacer sequence and PBS need be optimized. To maximize insertion efficiency by TwinPE, we recommend two rounds of atgRNA sequence optimization, testing variations in the spacer or PBS in each round. We recommend first optimizing the spacer sequence(s) by fixing the PBS and cloning atgRNAs containing different spacers targeting the genomic locus for insertion. Then, in a second round of optimization, the user can fix the optimal spacer sequence(s) identified in the first round of testing and try shortening and lengthening the PBS.

The choice of genomic locus to target with atgRNAs is an important one and depends on the intended use of PASTE. For some applications, such as in-frame protein tagging, the user already knows where they wish to direct the insertion and will be limited by the availability of protospacer sequences and space in which they can screen atgRNA combinations. For other applications, such as intron insertion for gene replacement, there is more flexibility for selection of target site, creating a large design space the user can navigate. To start, we generally recommend searching for protospacer sequences within an initial window of up to about 200bp, with the greater number of protospacers resulting in more atgRNAs that can be tested for high insertion efficiencies.

To design atgRNA spacer sequences, we recommend using Benchling’s CRISPR Design Tool to find protospacers close to the desired editing locus. The user can import the target genomic region into Benchling and select CRISPR > Design and analyze guides. Select single guide or paired guides depending on if the user wishes to insert attB sequences using a single atgRNA or TwinPE, respectively. Select the PAM sequence compatible with the Cas9 to be used in experiments (typically SpCas9, which has an NGG PAM sequence). If the user is flexible on the exact insertion site, it can be useful to search across several potential insertion sites to identify regions with a high number of protospacers. For insertion by TwinPE, we recommend that the user identifies pairs of spacers targeting the sense and antisense strand of DNA within a window of about 150 bp. If the user wishes to test nicking sgRNAs (PE3), use the CRISPR tool to identify spacers targeting the opposite strand of DNA approximately 60–100 bp downstream or upstream of the pegRNA cut site. Export the resulting spacer sequences to a .csv file.

Next, use previously published guidelines to design the PBS sequence¹⁵. We recommend beginning with a PBS length of 13 bp; in the second round of atgRNA optimisation, this can be varied between 8 and 15 bp. Web-based tools such as PrimeDesign⁴⁷, PE-Designer⁴⁸, pegIT⁴⁹ or pegFinder⁵⁰ can be used to assist with PBS design, although, we recommend the user to have a thorough understanding of the atgRNA architecture. We have published a software tool for atgRNA designs based on pooled screening of many different endogenous targets, named atgRANK (https://github.com/abugoot-lab/atgRNA_rank); this tool can also be used for designing atgRNAs for novel genomic sites²⁷.

For atgRNAs to be expressed from a U6 promoter, it is recommended that the most 5′ base is a guanine⁷. Append a 5′G to atgRNA sequences if there is not already one there. The user can either order the spacer, scaffold and template/PBS as separate oligos, anneal them, and clone by Golden Gate assembly as described previously⁵¹ or append 5′ and 3′ homology arms for Gibson assembly to the atgRNA sequences and order the atgRNAs as IDT e-block gene fragments. The latter is our preferred method as it is more convenient, however, it is more expensive. For nicking guides, due to their simpler architecture, we perform a PCR followed by Golden gate assembly into the target vector, which is cheaper and faster than ordering gene fragments.

Finally, when using PASTE multiplex insertion (see Box 2), the central dinucleotide of the attB to be inserted must match the central dinucleotide of the attP in the payload plasmid. The payload plasmids and atgRNA sequences we have provided have a GT central dinucleotide, which the user can change to other dinucleotides for multiplexing purposes. Based on our previous results^27,52, the GA, AG, AC, or CT dinucleotides are the most efficient and effective for orthogonal PASTE insertion.

Selection of prime editing system and PASTE architecture

While the first step of the PASTE process is always the insertion of the attB, there are a variety of different editors to choose from for that purpose. In particular, the choice between effectors for prime editing can have an impact on overall rates for certain targets.

The selection of a particular prime editing system for the application of PASTE will likely depend on the cell type and target of interest. For example, versions of prime editors have been developed for regions with less accessible chromatin structure⁵³. Several generations of prime editors have been developed to address challenges of the prime editing mechanism. The original prime editors (PE1, PE2, PE3) used either pegRNAs alone or in conjunction with a nicking guide to remove flap removal bias (PE3), an optimization that improves inclusion rates of the edit¹⁵. Later generations improved editing efficiency through modifications to the pegRNA architecture, mutagenesis (PE6), or suppression of host factors (PE4, PE5) that push the resolution of the editing event away from the desired product^20,22,54, as well as a variety of factors that that affect protein stability, expression, or localization. We recommend trying both PE3 and PE6, including PE6c and PE6d, as these systems vary in efficiency and evaluating all of them will help to maximize efficiency at a given target. Testing of different prime editing systems involves the cloning of these constructs into provided PASTE machinery plasmids to replace the PE2 prime editor. As an initial test, the user can deliver the prime editors and Bxb1 plasmids in trans to assess which prime editor version works best, before cloning into the PASTE machinery plasmid linked to the Bxb1.

Generally speaking, the choice between PE3-related systems (using a pegRNA and nicking guide, such as PEmax/PE6) and twinPE based systems will have the largest impact on experimental design. TwinPE approaches may result in the deletion of intervening sequences between the pegRNA targets depending on their placement, while PE3 can generally be designed to restore the surrounding sequence. The importance of this deletion will be dependent on the significance of the sequence around the PASTE target, such as maintenance of the reading frame within the coding DNA, as PE3 approaches would result in attL and attR scars around the inserted payload and twinPE may additionally delete bases. However, twinPE is generally more efficient for attB insertion and can therefore result in higher overall rates. This balance should be evaluated depending on target and application.

Similarly, multiple generations of PASTE can be utilized depending on the desired edit. The current generation PASTEv3 and PASTEv4 employ Bxb1 and B.cereus integrases respectively, while PASTEv1 employs a different architecture with a self-cleaving 2A linker connecting the integrase allowing for separate expression of the Bxb1 integrase. There are locus dependent applications where the overall editing rates of each of these strategies may be highest. Generally, the PASTEv3 or PASTEv1 systems are good starting points to begin evaluating the atgRNA and nicking guide designs from.

Payload design considerations

The architecture of a PASTE payload is quite simple, as it requires only the DNA sequence to be inserted and an attP sequence, which is usually placed upstream of the insert DNA. We have provided a plasmid which contains an EGFP downstream of an attP sequence (pDY0181, Addgene ID: 179115). This plasmid is compatible with the NGS/ddPCR primers and probes listed in Table 1 and can be used to optimize many of the experiments described in this protocol. While some background GFP expression from the episomal (i.e. uninserted) plasmid may be observed in transfected cells, this payload lacks a promoter sequence to drive GFP expression in edited cells. Therefore, integration efficiency should be quantified by NGS or ddPCR, using the primers and probes provided in Table 1, rather than by GFP fluorescence. We have also provided a protocol for cloning a custom PASTE payload. When designing a payload plasmid, there are several modifications the user can make to tailor the plasmid to their experiment. The major considerations for a payload are as follows:

Table 1.

Sequencing and cloning oligos

NGS primers round 1 (5’ -> 3’)				Step
	Handle for round 2	Primer binding sequence	complete sequence
NOLC1_NGS_F1 (PD7968)	ACACTCTTTCCCTACACGACGCTCTTCCGATCTC	AATGACGTAACACAGGCCCGC	ACACTCTTTCCCTACACGACGCTCTTCCGATCTCAATGACGTAACACAGGCCCGC	47, 66
NOLC1_NGS_F2 (PD7969)	ACACTCTTTCCCTACACGACGCTCTTCCGATCTAC	AATGACGTAACACAGGCCCGC	ACACTCTTTCCCTACACGACGCTCTTCCGATCTACAATGACGTAACACAGGCCCGC	47, 66
NOLC1_NGS_F3 (PD7970)	ACACTCTTTCCCTACACGACGCTCTTCCGATCTGAC	AATGACGTAACACAGGCCCGC	ACACTCTTTCCCTACACGACGCTCTTCCGATCTGACAATGACGTAACACAGGCCCGC	47, 66
NOLC1_NGS_F4 (PD7971)	ACACTCTTTCCCTACACGACGCTCTTCCGATCTTGAC	AATGACGTAACACAGGCCCGC	ACACTCTTTCCCTACACGACGCTCTTCCGATCTTGACAATGACGTAACACAGGCCCGC	47, 66
NOLC1_NGS_R (PD7972)	GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCT	CGAGCACGAGGGGATACAGGTC	GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCTCGAGCACGAGGGGATACAGGTC	47, 66
pDY0181_GFP_insert_NGS_R (PD6893)	GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCT	GAACTCCACGCCGTTCAGGG	GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCTGAACTCCACGCCGTTCAGGG	47, 66
ACTB_NGS_F1 (PD0966)	ACACTCTTTCCCTACACGACGCTCTTCCGATCT	CCGACCTCGGCTCACAGCG	ACACTCTTTCCCTACACGACGCTCTTCCGATCTCCTGGCGGCCTAAGGACTCG	47, 66
ACTB_NGS_F2 (PD0967)	ACACTCTTTCCCTACACGACGCTCTTCCGATCTA	CCGACCTCGGCTCACAGCG	ACACTCTTTCCCTACACGACGCTCTTCCGATCTCCTGGCGGCCTAAGGACTCG	47, 66
ACTB_NGS_F3 (PD0968)	ACACTCTTTCCCTACACGACGCTCTTCCGATCTGA	CCGACCTCGGCTCACAGCG	ACACTCTTTCCCTACACGACGCTCTTCCGATCTCCTGGCGGCCTAAGGACTCG	47, 66
ACTB_NGS_F4 (PD0969)	ACACTCTTTCCCTACACGACGCTCTTCCGATCTTGA	CCGACCTCGGCTCACAGCG	ACACTCTTTCCCTACACGACGCTCTTCCGATCTCCTGGCGG	47, 66
			CCTAAGGACTCG
ACTB_NGS_F5 (PD0970)	ACACTCTTTCCCTACACGACGCTCTTCCGATCTCTGA	CCGACCTCGGCTCACAGCG	ACACTCTTTCCCTACACGACGCTCTTCCGATCTCCTGGCGGCCTAAGGACTCG	47, 66
ACTB_NGS_F6 (PD0971)	ACACTCTTTCCCTACACGACGCTCTTCCGATCTACTGA	CCGACCTCGGCTCACAGCG	ACACTCTTTCCCTACACGACGCTCTTCCGATCTCCTGGCGGCCTAAGGACTCG	47, 66
ACTB_NGS_F7 (PD0972)	ACACTCTTTCCCTACACGACGCTCTTCCGATCTTACTGA	CCGACCTCGGCTCACAGCG	ACACTCTTTCCCTACACGACGCTCTTCCGATCTCCTGGCGGCCTAAGGACTCG	47, 66
ACTB_NGS_F8 (PD0973)	ACACTCTTTCCCTACACGACGCTCTTCCGATCTGTACTGA	CCGACCTCGGCTCACAGCG	ACACTCTTTCCCTACACGACGCTCTTCCGATCTCCTGGCGGCCTAAGGACTCG	47, 66
ACTB_NGS_R (FP0952)	GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCT	CCACCCAGCCAGCTCCC	GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCTCCACCCAGCCAGCTCCC	47, 66
ddPCR primers and probe (5’ -> 3’)
NOLC1_ddPCR_F (PD1022)	TGGAGCCCACCCTTTCCGT			84
ACTB_ddPCR_F (PD2053)	CCCGGCTTCCTTTGTCC			84
attL_ddPCR_probe (PD1219)	/56-FAM/CC GGC TTG T/ZEN/C GAC GAC GGC G/3IABkFQ/			84
Cloning primers (5’ -> 3’)
Payload-Gibson-F	ACCACCGCGGTCTCAGTGGTGTACGGTACAAACCCAGCTACCGGTNNNNNNNNNNNNN			24
Payload-Gibson-R	AATAAAGCATTTTTTTCACTGCATTCTAGTTGTGGTTTGTCTAGATCANNNNNNNNNNNNN			24
Nicking guide PCR F (replace N with spacer sequence)	ATggtctcgcaccgNNNNNNNNNNNNNNNNNNNNgttttagagctagaaatagcaagttaaaataaggct			13
Nicking guide PCR R (universal)	atggtctctaaaagcaccgactcggtgccac			13
atgRNA (5’ -> 3’)	5’ overhang	atgRNA sequence	3’ overhang
pDY2259 hNOLC1 atgRNA Paired Guide 1 (Addgene ID: 220989)	tcttgggtagtttgcagttttaaaattatgttttaaaatggactatcatatgcttaccgtaacttgaaagtatttcgatttc	GCGTATTGCCTGGAGGATGGgttttagagctagaaatagcaagttaa	tttttttaagcttgggccgctcgaggtacctctctacatatgac	37, 64
	ttggctttatatatcttgtggaaaggacgaaacacc	aataaggctagtccgttatcaacttgaaaaagtggcaccgagtcggtgcccggatgatcctgacgacggagaccgccgtcgtcgacaagccggcctcctccagg	atgtgagcaaaaggccag
pDY2260 hNOLC1 atgRNA Paired Guide 2 (Addgene ID: 220990)	tcttgggtagtttgcagttttaaaattatgttttaaaatggactatcatatgcttaccgtaacttgaaagtatttcgatttcttggctttatatatcttgtggaaaggacgaaacacc	gTATTGGCCACCTCTGAGAGTgttttagagctagaaatagcaagttaaaataaggctagtccgttatcaacttgaaaaagtggcaccgagtcggtgcggccggcttgtcgacgacggcggtctccgtcgtcaggatcatccggctcagaggt	tttttttaagcttgggccgctcgaggtacctctctacatatgacatgtgagcaaaaggccag	37, 64
pDY2261 hACTB atgRNA Paired Guide 1 (Addgene ID: 220991)	tcttgggtagtttgcagttttaaaattatgttttaaaatggactatcatatgcttaccgtaacttgaaagtatttcgatttcttggctttatatatcttgtggaaaggacgaaacacc	GCTATTCTCGCAGCTCACCAgttttagagctagaaatagcaagttaaaataaggctagtccgttatcaacttgaaaaagtggcaccgagtcggtgcATGATCCTGACGACGGAGACCGCCGTCGTCGACAAGCCtgagctgcgagaa	tttttttaagcttgggccgctcgaggtacctctctacatatgacatgtgagcaaaaggccag	37, 64
pDY2262 hACTB atgRNA Paired Guide 2 (Addgene ID: 220992)	tcttgggtagtttgcagttttaaaattatgttttaaaatggactatcatatgcttaccgtaacttgaaagtatttcgatttcttggctttatatatcttgtggaaaggacgaaacacc	GGAGGGGAAGACGGCCCGGGgttttagagctagaaatagcaagttaaaataaggctagtccgttatcaacttgaaaaagtggcaccgagtcggtgcggcttgtcgacgacggcggtctccgtcgtcaggatcatgggccgtcttccc	tttttttaagcttgggccgctcgaggtacctctctacatatgacatgtgagcaaaaggccag	37, 64
Sanger sequencing primers (5’ -> 3’)
hU6-F	GAGGGCCTATTTCCCATGATT			12
LKO.1 5’	GACTATCATATGCTTACCGT			12
pCI-rev	GCAATAGCATCACAAATTTCAC			29
Synthetic RNA sequences
mNOLC1 twinPE pegRNA 1	mA*mG*mU*UAAGGAGGCGAGGGCUGGUUUUAGAGCUAGAAAUAGCAAGUUAAAAUAAGGCUAGUCCGUUAUCAACUUGAAAAAGUGGCACCGAGUCGGUGCGGCUUGUCGACGACGGCGGUCUCCGUCGUCAGGAUCAUCCCUCGCmC*mU*mC			37, 64
mNOLC1 twinPE pegRNA 2	mA*mC*mA*CCGAGACCUCCAGCCUGGUUUUAGAGCUAGAAAUAGCAAGUUAAAAUAAGGCUAGUCCGUUAUCAACUUGAAAAAGUGGCACCGAGUCGGUGCAUGAUCCUGACGACGGAGACCGCCGUCGUCGACAAGCCGCUGGAGmG*mU*mC			37, 64
IVT primers (5’ -> 3’)
PASTE IVT F (PD3771)	agataTAATACGACTCACTATAGGGAGAgccgccaccatgaaacgga			Box 3
PASTE IVT R (PD3762)	cga ggc tga tca gcg ggt tta aac			Box 3

First, the size of the insert, which affects possible formats for delivery. While transient transfection is useful for screening, different delivery vehicles can accommodate different maximum cargo sizes. When using an AAV to deliver the cargo, a sequence up to approximately 4.7 kb can be delivered. If using Adenovirus to deliver the cargo, the theoretical maximum cargo size increases to at least ~36 kb if using a helper-dependent adenovirus design that has removed most of the endogenous viral sequences, allowing for increased capacity^55,56.

Second, the frame of the insert, which is particularly important for applications such as in-frame tagging or insertion of a region of amino acids, must be considered when designing the template sequence. The frame can be manipulated by adding bases in the cargo sequence between the attP (which becomes the attL when recombined) and the feature to be translated (Fig. 4a). When designing in-frame tagging constructs, be sure to consider the frame of the resulting insert relative to the attL sequence that recombines in the genome.

Figure 4: Modifying the frame of the PASTE insert and generation of NGS barcodes or atgRNA crosses.

a) This schematic shows with sequence resolution how to adjust the frame of the inserted PASTE payload. Briefly, the insertion of 1 or 2 bases within the payload between the 3′ end of the attP (which becomes the 3′ end of the attL after integration) and the GOI or other cargo will shift the amino acid frame. The attB and attP are highlighted dark orange and cyan, while the payload position is shown in green. 1 or 2 bases can be inserted in the orange highlighted position labeled “NN” depending on the desired frameshift. b) Schematic describing the creation of up to 96 unique crosses of second-round barcoding sets using Illumina forward and reverse primers, or combinations of nicking and atgRNA as required.

Third, whether the user wants only the payload to be inserted or is content with the payload and the rest of the payload plasmid being inserted into the genome is an important design parameter. When using a single attB/attP combination and a full-length payload plasmid (as is described in this protocol), Bxb1 integrase will insert the entire plasmid into the attB site. If the user wants only the payload sequence and not the rest of the plasmid required for bacterial propagation (e.g. antibiotic resistance gene, ori, etc.), one option is to produce a minicircle⁵⁷ which can exclude all but the sequence to be inserted from the payload plasmid and is compatible with the transfection and readouts described in this protocol. Additionally, we have found PCR products or annealed oligos to be amenable for insertion with no detectable indels formed due to the non-circular nature of the insert. Lastly, if using viral vectors (see Box 1), such as AAV or adenovirus, one can avoid inserting other bacterial sequences on plasmids, but it is important to consider that the viral sequences on your vector (e.g., the AAV ITRs) will be inserted and to evaluate the effects of these sequences on your host genome and cell.

Finally, whether any additional sequence features need to be added should be considered. For example, the addition of stop or start codons (depending on payload), or elements such as a splice acceptor (if the inserted sequence should splice with an upstream exon) depending on the expected application of the inserted product.

Readouts for integration efficiency

To readout insertion efficiencies of attB sites and/or payload insertion, we have provided protocols for a 2-primer NGS assay (for attB site insertion), left-junction 3-primer NGS assay (for payload insertion) and left-junction ddPCR (for payload insertion) (Fig. 5). Our NGS library prep protocols use two rounds of PCR amplification. The first round uses target-specific primers that land approximately 100–150 bp upstream and downstream of the insertion site, with overhangs that add handles for the second round of PCR. The primers used in the second round add Illumina adapters and barcodes required for multiplexed sequencing on an Illumina machine. For the user’s convenience, we have provided the sequences of validated primers and probes we use to assess attB and/or insertion of an EGFP payload at the NOLC1 and ACTB loci in Table 1.

Figure 5: Overview of design for NGS and ddPCR assays.

An outline showing the placement of forward and reverse primers or probes for reading out PASTE experiments. The two-primer NGS assay can be used to measure attB insertion by the prime editor into the genomic target, by placing primers flanking the desired insertion site. The three-primer NGS assay adds a primer that binds within the insert if PASTE integration is successful. Finally, the ddPCR assay uses a reverse primer that binds within the payload, as well as a probe, so that amplification and fluorescence occurs only with successful PASTE integration.

For 2-primer assays of attB site insertion, this involves simply designing a forward primer pool and a reverse primer that amplify the locus of interest and adding the round 2 PCR handles as overhangs. The forward primer pool contains primers that are almost identical but contain index barcodes and different stagger sequences for sample multiplexing and to balance representation of nucleotides in each sequencing cycle (known as diversity) on the Illumina flow cell. For 3-primer NGS assays, we use a forward primer pool and two reverse primers: one complementary to the 5′ side of the payload plasmid insertion site (this amplifies the insert) and the other reverse primer complementary to the genomic locus downstream of the insertion site (this amplifies the non-inserted site). This results in two amplicons being produced: one consisting of the unedited genomic locus and one consisting of the genomic locus post payload insertion. After NGS, insertion efficiency is calculated as the ratio of the wildtype amplicon to edited amplicons. The advantages of this method are its ease and relative low cost. However, it assumes that the PCR amplifies both the inserted and non-inserted amplicons with equal efficiency. Any PCR bias towards one of the amplicons will result in an inaccurate estimation of the insertion efficiency in the genome. Therefore, steps should be taken to minimize any potential PCR bias (such as ensuring both amplicons are the same size and minimizing PCR cycles) and standards may need to be used when establishing a new assay to confirm accurate estimation of amplicon ratios. Standards can either be synthesized DNA fragments or genomic clones containing the insert. While genomic clones are generally considered the gold standard, in practice we have found synthesized DNA to be just as accurate for standard generation.

Droplet digital PCR (ddPCR) is another assay for quantifying PASTE insertion efficiencies that is orthogonal to the NGS methods described already. An advantage of ddPCR is that it does not have the same risk of amplification bias as with the NGS assay and is considered a more robust and accurate method of assessing insertion efficiencies. However, compared to NGS, the reagents for ddPCR are more expensive. Also, it requires more specialized equipment and we find it to be generally more challenging to optimize. The reverse primer and probe that we have provided are compatible with integrating the EGFP cargo with attP site pDY0181 (Addgene ID: 179115) and the provided ACTB or NOLC1 forward primer (see Table 1) or, if targeting a new locus, with a custom forward primer. For custom payloads cloned into the payload acceptor with attP site (PASTE payload acceptor with attP site pDY2087 (Addgene ID: 219859)), the user will need to design a custom reverse primer.

Importantly, the assays described in this protocol infer payload insertion efficiency by using assays that target the left junction of the insert only (Fig. 5). As Illumina sequencing can only sequence short amplicons and ddPCR does not give sequence-level information, the full sequence of a long insert must be assessed by other methods. To this end, the user can employ long-read sequencing methods such as Oxford Nanopore sequencing or PacBio HiFi sequencing.

Delivery of PASTE components

Successful PASTE insertion requires efficient delivery of the PASTE components and we recommend performing most of the optimization experiments in easy to handle cell lines such as HEK293FT. To transfect cells in culture, we typically utilize lipofectamine 3000 to deliver plasmid DNA by transfection. Other methods for transfection such as Trans-IT or polyethylenimine (PEI) may be compatible depending on the cell line of interest or may have different suitability for different sized cargoes. For delivery of PASTE components as RNA instead of plasmid DNA, we recommend ordering the atgRNAs as synthetic RNAs from Synthego as an EZ-sgRNA or from IDT as Alt-R sgRNAs, with the first and last 3 bases containing 2’ O-Methyl and phosphorothioate bond modifications for end protection, which enhance RNA stability⁵⁸ (see Table 1 for an example). We have also provided a protocol for in-vitro transcription (IVT) of the PASTE machinery (Box 3). We find the inclusion of 5′ and 3′ UTRs in the mRNA to be helpful. Please note that, due to cargo constraints of AAV vectors, dual AAV delivery is required to deliver the PASTE components (protein machinery and guides) to cells or in vivo, not including the template, which would be a third vector. We have also used adenovirus to deliver the PASTE components and a single vector design is possible, including the proteins, guides, and template, if using a helper-dependent adenovirus approach that allows for ~36 kb of DNA capacity (see Box 1). The user should also note that lentivirus delivery of the payload is theoretically possible, as lentivirus has an RNA genome which is reverse transcribed to cDNA, but we have not tested this ourselves. Integration payloads can also by generated by PCR; however, Bxb1 integration of a linear payload will result in a double strand break that is resolved by endogenous DNA repair machinery.

(BOX 3). Production of in vitro transcribed PASTE mRNA (optional) (Timing = 1d).

Critical. RNAs are susceptible to degradation by RNases. Therefore, we recommend the user to practice good RNase-free technique. Before starting, wipe down surfaces and pipettes with RNase away or RNaseZap and work with tips and tubes certified as RNase free.

• 1Preparation of DNA template. DNA templates for IVT must be linear and must also contain a 5′ T7 promoter sequence. Our PASTE plasmids already contain a 5′ T7 promoter, which means that a restriction digest with an enzyme that cuts uniquely 3′ of the PASTE ORFs (e.g. HindIII) followed by gel purification is sufficient to prepare the template. However, we generally prefer to generate the linear DNA template by PCR using primers containing a 5′ T7 promoter sequence as an overhang. To do so, set up a PCR using the following recipe, scaling up as appropriate:

Component	Amount per reaction (μL)	Final concentration
NebNext High-Fidelity 2x PCR Master Mix (NEB)	12	1x
IVT Forward Primer (50uM) PD3771 (Table 1)	0.2	0.2μM
IVT Reverse Primer (50uM) PD3762 (Table 1)	0.2	0.2μM
Plasmid template (e.g. pDY0216) 10ng/μL	1	2ng/μL
UltraPure water	To 50μL

• 2Incubate the PCRs in a thermal cycler under the following conditions:

Cycle number	Denature	Anneal	Extend
1	98 °C, 30 s
2 – 33	98 °C, 15 s	63 °C, 30 s	72 °C, 4 min
34			72 °C, 5 min

• 3Check a few microliters of the reaction on a 2% e-gel.
• 4Purify the reactions. If there is a clean ~8kb band, proceed to PCR purification by column, for example with Qiaquick PCR purification columns (Qiagen). If there are multiple bands seen on the gel, we recommend excising the 8kb band from the gel and purifying using Monarch Gel Purification kit (NEB).
• 5Prepare IVT reactions. Set up the following master mix, scaling up as appropriate.

Component	Amount per reaction (μL)	Final concentration
ARCA/NTP 2x Mix (NEB)	10	1x
Template DNA (from Step 4 of this Box) 250ng/μL	4	50ng/μL
Pseudouridine-5′-triphosphate 100mM (Trilink)	0.25	1.25mM
T7 RNA polymerase mix (NEB)	2	1x
Murine RNase inhibitor (NEB)	0.5	1000U/μL
UltraPure water	To 20μL

Critical step: This Master mix recipe contains two RNA modifications: the ARCA cap and pseudouridine, which improve RNA stability and reduce innate immune responses, improving cell health and editing efficiency. However, if the user prefers to omit these modifications, simply remove the pseudo-U and use the NTP mix from NEB. Additional modified nucleotides can be used to further reduce cellular innate immune responses. This recipe produces RNAs containing a mixture of uridine and pseudo-uridine; if the user wishes there to be only pseudo-U, simply use a mastermix where the nucleotides are not premixed and substitute the uridine for psuedo-uridine.

• 6Mix the samples well by pipetting (do not vortex). Incubate the above reaction at 37°C for 30 mins.
• 7Degrade plasmid template. Add 2μL of DNase I to each reaction and mix well by pipetting. Do not vortex.
• 8Incubate the reactions for another 15 minutes at 37°C.
• 9Place the reactions on ice and check 1 μL of the reactions on a 2% e-gel to confirm successful transcription. The user may also wish to check RNA integrity using a Bioanalyzer or similar, though we usually omit this.

  ?Troubleshooting

• 10(optional) Poly(A) tailing of mRNAs. While the mRNAs will be functional without a tail, we recommend poly-A tailing for improved mRNA stability. Working on ice, set up the following master mix, scaling up as appropriate. Alternatively, skip to step 13.

Component	Amount per reaction (μL)	Final concentration
IVT reaction (from Step 9 of this Box)	22
10x Poly(A) Polymerase reaction buffer	5	1x
Poly(A) Polymerase	5
UltraPure water	To 50μL

• 11Mix the reactions well by pipetting. Do not vortex.
• 12Incubate the reactions at 37°C for 30 minutes.
• 13Purify the reactions using Monarch RNA Cleanup Kit (NEB), following the manufacturer’s protocol.
• 14Using a Nanodrop UV spectrophotometer, analyse 1μL of each sample. For RNA, the 260/280 values should be above 2. Optionally, run 1μL of RNA (diluted in 9 μL water) on 1 lane of a 2% e-gel to confirm an RNA at the correct size and no degradation.

  Pause point. RNA samples may be stored at −80°C for at least 2 weeks. Freeze-thaw cycles should be avoided to preserve RNA integrity.

Experimental controls

For all PASTE editing experiments, it is critical to include a negative control, which contains all components of the test samples (payload plasmid, atgRNAs) and either omit the PASTE machinery plasmid or replace it with an appropriate negative control plasmid. We typically use a plasmid expressing Bxb1 only (Addgene ID: 51271). These controls are critical for sequencing experiments, as they contribute a reference amplicon against which test samples can be compared, controlling for the presence of mutations (e.g. SNPs, indels) that may occur in the cell line of choice. Additionally, readout experiments are confounded if PCR misamplification occurs off of one of the transfected plasmids, which may occur due to the reverse primer hybridizing to the payload plasmid. In our experience, PCR mispriming is more likely to occur during ddPCR readouts rather than NGS. Therefore, it is critical that the negative control contains the payload and atgRNA plasmids, rather than simply using an untransfected control. If editing a new genomic locus or using new PASTE components, we recommend including a positive control in the experiment, which could be previously validated atgRNAs and payload, including those provided in this protocol. We also recommend that all transfections within the same experiment be performed at the same time and on cells that were seeded in parallel to control for variables such as transfection efficiency and cell density, which can greatly affect editing efficiency.

Assessing off target integrations

PASTE applications, including those with therapeutic potential, may require quantification of off-target integrations at unintended genomic loci. There are several sequencing based methods in the literature which can be applied to map PASTE integrations across the genome. We have had success in our lab using LAM-HTGTS⁵⁹, adapting the protocol by expanding the transfected cells over a period of two weeks (to deplete the payload plasmid) and utilizing a primer that amplifies from the integrated plasmid into the surrounding genome. We have also successfully applied the tagmentation-based UdiTas approach⁶⁰, with similar modifications. We expect that other approaches described in the literature can be similarly adapted to mapping PASTE integrations in the genome, including in cell approaches such as GUIDE-seq⁶¹, or in vitro approaches such as CHANGE-seq⁶².

MATERIALS

Biological Materials

• HEK293FT cells (Invitrogen cat. R70007) (RRID:CVCL_6911, https://scicrunch.org/resolver/RRID:CVCL_6911)

!Caution. All cell lines should be regularly tested for mycoplasma

Reagents

Plasmids

• PASTE editor XTEN linker pDY1052 (Addgene ID: 179105)

• PASTE editor P2A linker pDY0216 (Addgene ID: 179104)

• pCAG-NLS-HA-Bxb1 (Addgene ID: 51271)

• EGFP cargo with attP site pDY0181 (Addgene ID: 179115)

• PASTE payload acceptor with attP site pDY2087 (Addgene ID: 219859)

• NOLC1 atgRNA pDY2250 (Addgene ID: 219860)

• NOLC1 nicking guide pDY2251 (Addgene ID: 219861)

• pDY2259 hNOLC1 atgRNA Paired Guide 1 (Addgene ID: 220989)

• pDY2260 hNOLC1 atgRNA Paired Guide 2 (Addgene ID: 220990)

• pDY2261 hACTB atgRNA Paired Guide 1 (Addgene ID: 220991)

• pDY2262 hACTB atgRNA Paired Guide 2 (Addgene ID: 220992)

• pU6-pegRNA-GG-acceptor (Addgene ID: 132777)

• pDY0209 ACTB nicking guide (Addgene ID: 179109)

Cloning

• PCR-grade H₂O (Invitrogen cat. 10977015)

• NEBNext High-Fidelity 2x PCR Master Mix (NEB cat. M0541L)

• NEBuilder HiFi DNA Assembly Master Mix (NEB cat. E2621X)

• 1 Kb Plus DNA Ladder (Invitrogen cat. 10787026)

• One Shot Stbl3 Chemically Competent E. coli (Invitrogen cat. C737303)

• Monarch DNA Gel Extraction Kit (NEB cat. T1020L)

• E-gel EX Gel, 2% (Invitrogen cat. G402002)

• EconoSpin 96 well filter plate (Fisher Scientific cat. NC0231536)

• Qiagen buffer P1 (Qiagen cat. 19051)

• Qiagen buffer P2 (Qiagen cat. 19052)

• Qiagen buffer N3 (Qiagen cat. 19064)

• Nucleobond Xtra Midi Kit (Machery Nagl cat. 740410.50)

• Terrific Broth (Granulated) (Fisher Scientific cat. BP9728–500)

• SOC outgrowth media (NEB cat. B9020SVIAL)

• LB Agar (VWR cat. J7851-A1)

• Ampicillin sodium salt (Sigma-Aldrich cat. A9518)

• FastDigest XbaI (Thermo Fisher Scientific cat. FD0684)

• FastDigest BshTI (Thermo Fisher Scientific cat. FD1464)

• FastDigest Eco47III (Thermo Fisher Scientific cat. FD0324)

NGS

• Quick Extract Buffer (Lucigen cat. QE0901L)

• MiSeq Reagent Micro Kit v2 (300 cycles) (Illumina cat. MS-103–1002)

• MiSeq Reagent Nano Kit v2 (300 cycles) (Illumina cat. MS-103–1001)

• PhiX Control Kit v3 (Illumina cat. FC-110–3001)

• Sodium hydroxide solution (Sigma-Aldrich cat. S8263)

• Amplification Primers (Integrated DNA Technologies) see Table 1

• Qubit 1x dsDNA HS Assay Kit (Thermo Fisher cat. Q33231)

ddPCR

• Magnetic Beads

• Sera-Mag SpeedBead Carboxylate-Modified Magnetic Particles (Hydrophobic) (Cytiva cat. 65152105050250)

• Droplet generation oil for probes (Bio-Rad cat. 1864005)

• Droplet generation oil for EvaGreen (Bio-Rad cat. 1863005)

• ddPCR droplet reader oil (Bio-Rad cat. 186–3004)

• ddPCR 96-well plates (Bio-Rad cat. 12001925)

• ddPCR Buffer control for probes (Bio-Rad cat. 1863052)

• ddPCR Copy Number Assay RPP30 Human HEX (Bio-Rad cat. 10031245)

• QX200 Droplet Reader (Bio-Rad cat. 1864003)

• Supermix for probes (Bio-Rad cat. 1863024) or Supermix for EvaGreen (Bio-Rad cat. 1864034)

• Consumables kit for droplet generation (Bio-Rad cat. 1863008)

• Primers – see Table 1

Mammalian cell culture

• DMEM, high glucose (Gibco cat. 10569010)

• Penicillin/streptomycin (Gibco cat. 15140163)

• Fetal bovine serum (Gibco cat. 16000044)

  • !Critical. FBS should be aliquoted and stored at −20oC.

• TrypLE express enzyme (Gibco cat. 12604039)

• Phosphate-buffered saline, Dulbecco’s formula (Thermo Fisher Scientific cat. J678-2.K2)

• Lipofectamine 3000 transfection reagent (Invitrogen cat. L300000150)

• Opti-MEM (Thermo Fisher Scientific cat. 31985088)

• MycoAlert Plus (Lonza cat. LT07–710)

Equipment

• 8-well strip tubes for PCR (VWR cat. 76102–482)

• 96-well PCR plates (VWR cat. 47743–953)

• Axygen PCR aluminium sealing film (Corning cat. PCR-AS-200)

• 1.5mL microcentrifuge tubes (VWR cat. 10011–744)

• Falcon 50mL centrifuge tubes (VWR cat. 21008–940)

• Falcon 15mL centrifuge tubes (VWR cat. 734–0451)

• DynaMag-96 side skirted magnet (Invitrogen cat. 12027)

• Thermocycler - any

• Plate seal - clear, porous

• Vacuum filter system TC

• Serological pipettes

• Cell counter

• TC dishes

• TC plates

• Tabletop centrifuge

• Gel developer

• Droplet Digital PCR system (Bio-Rad QX200 or QX200 AutoDG droplet digital PCR system)

• Pipet tips for the AutoDG system (Bio-Rad cat. 1864120)

• E-gel electrophoresis device (Thermo Fisher cat. G8100)

• Safe imager viewing glasses (Thermo Fisher cat. S37103)

• Qubit 4 fluorometer (Thermo Fisher cat. Q33238)

• Qubit assay tubes (Thermo Fisher cat. Q33237)

• Nanodrop spectrophotometer (Thermo Fisher cat. ND-2000C)

• MiSeq (Illumina cat. SY-410–1003)

• Python 3+ (https://www.python.org/)

• CRISPResso2 (https://github.com/pinellolab/CRISPResso2)⁶³

REAGENT SETUP

• DMEM/10 % FBS media (D10). For culturing HEK293FTs, prepare D10 medium by supplementing DMEM with 10% (vol/vol) FBS and 1x penicillin/streptomycin. Store the medium at 4°C for up to 2 months.

• Magnetic beads. For purification of genomic DNA after extraction using QE buffer, aliquot 10mL of speed beads into a 50mL falcon tube and place the tube on a magnetic separator. Wait until the beads have fully separated from solution, then remove the supernatant and discard. Remove the tube from the magnetic separator and resuspend the beads in 10mL DI water. Place the tube back on the separator and wait until separated. Repeat this wash for a total of 2x DI water washes. Prepare a bead buffer containing 100mL of 5M NaCL, 5mL of 1M Tris-HCl, 1mL of 500mM EDTA and 90g of PEG-8000. Invert this buffer 10x and fill to 500mL with DI water. Sterile filter this buffer. Resuspend the pelleted beads in 10mL of the prepared bead buffer and vortex to mix. Add the bead slurry to the remaining ~490mL of bead buffer and swirl the bottle to distribute. Aliquot the beads in 50mL falcon tubes and freeze at −20°C. Frozen beads can be stored indefinitely at −20°C and for at least 6 months at 4°C.

• NGS round 2 PCR primer mixes. Order the 8 forward and 12 reverse primers (see sequences in Table 2) and array them by column (forward primers) or by row (reverse primers) in a 96-well plate. Dilute them to 2.5μM in water. In a separate 96 well PCR, create unique combinations of forward and reverse primers by multi-channeling forward primers into each row and reverse primers into each column of a 96 well plate, as shown in Fig. 4b.

Table 2.

Second round NGS barcode primers

NGS round 2 PCR forward primers		Step
NGS_96_F_A1	AATGATACGGCGACCACCGAGATCTACACAAGTAGAGACACTCTTTCCCTACACGACGCTCTTCC	53, 67
NGS_96_F_A2	AATGATACGGCGACCACCGAGATCTACACCATGCTTAACACTCTTTCCCTACACGACGCTCTTCC	53, 67
NGS_96_F_A3	AATGATACGGCGACCACCGAGATCTACACGCACATCTACACTCTTTCCCTACACGACGCTCTTCC	53, 67
NGS_96_F_A4	AATGATACGGCGACCACCGAGATCTACACTGCTCGACACACTCTTTCCCTACACGACGCTCTTCC	53, 67
NGS_96_F_A5	AATGATACGGCGACCACCGAGATCTACACAGCAATTCACACTCTTTCCCTACACGACGCTCTTCC	53, 67
NGS_96_F_A6	AATGATACGGCGACCACCGAGATCTACACAGTTGCTTACACTCTTTCCCTACACGACGCTCTTCC	53, 67
NGS_96_F_A7	AATGATACGGCGACCACCGAGATCTACACCCAGTTAGACACTCTTTCCCTACACGACGCTCTTCC	53, 67
NGS_96_F_A8	AATGATACGGCGACCACCGAGATCTACACTTGAGCCTACACTCTTTCCCTACACGACGCTCTTCC	53, 67
NGS_96_F_A9	AATGATACGGCGACCACCGAGATCTACACACACGATCACACTCTTTCCCTACACGACGCTCTTCC	53, 67
NGS_96_F_A10	AATGATACGGCGACCACCGAGATCTACACGGTCCAGAACACTCTTTCCCTACACGACGCTCTTCC	53, 67
NGS_96_F_A11	AATGATACGGCGACCACCGAGATCTACACGTATAACAACACTCTTTCCCTACACGACGCTCTTCC	53, 67
NGS_96_F_A12	AATGATACGGCGACCACCGAGATCTACACTTCGCTGAACACTCTTTCCCTACACGACGCTCTTCC	53, 67
NGS round 2 PCR reverse primers:
NGS_96_R_E1	CAAGCAGAAGACGGCATACGAGATCATGATCGGTGACTGGAGTTCAGACGTGTGCTCTTC	53, 67
NGS_96_R_E2	CAAGCAGAAGACGGCATACGAGATAGGATCTAGTGACTGGAGTTCAGACGTGTGCTCTTC	53, 67
NGS_96_R_E3	CAAGCAGAAGACGGCATACGAGATGACAGTAAGTGACTGGAGTTCAGACGTGTGCTCTTC	53, 67
NGS_96_R_E4	CAAGCAGAAGACGGCATACGAGATCCTATGCCGTGACTGGAGTTCAGACGTGTGCTCTTC	53, 67
NGS_96_R_E5	CAAGCAGAAGACGGCATACGAGATTCGCCTTGGTGACTGGAGTTCAGACGTGTGCTCTTC	53, 67
NGS_96_R_E6	CAAGCAGAAGACGGCATACGAGATATAGCGTCGTGACTGGAGTTCAGACGTGTGCTCTTC	53, 67
NGS_96_R_E7	CAAGCAGAAGACGGCATACGAGATGAAGAAGTGTGACTGGAGTTCAGACGTGTGCTCTTC	53, 67
NGS_96_R_E8	CAAGCAGAAGACGGCATACGAGATATTCTAGGGTGACTGGAGTTCAGACGTGTGCTCTTC	53, 67

PROCEDURE

Cloning atgRNAs (Timing = 4d)

Digest acceptor plasmid.

• 1Set up a restriction digest containing the atgRNA plasmid acceptor (pU6-pegRNA-GG-acceptor, Addgene #132777), as follows. Scale up depending on the anticipated amount of digested plasmid needed (we typically expect 30–40% recovery after purification).

Component	Amount per reaction (μL)	Final concentration
pU6-pegRNA-GG-acceptor, (500ng/μl)	6	150ng/μl
FastDigest Eco31l	1
FastDigest buffer (10x)	2	1x
UltraPure water	11

• 2Incubate the digestion in a thermocycler for 1 hour at 37 °C.
• 3Dilute the digested product 1:4 with 80μL of water, then run 20μL per lane on a 2% E-Gel EX. Purify the band at about 2200bp using a Monarch DNA Gel Extraction Kit, following the manufacturer’s protocol.

Gibson assembly of atgRNAs.

• 4Set up a reaction containing the digested atgRNA plasmid acceptor from Step 2, and Gibson assembly master mix as follows and scale up depending on the number of atgRNAs to be cloned:

Component	Amount per reaction (μL)	Final concentration
Digested vector (50ng/μl)	1	10ng/μl
NEBuilder HiFi DNA Assembly Master Mix (2x)	2.5	1x
UltraPure water	0.5

• 5Using a multichannel pipette, distribute 4μL of the master mix per well of a 96-well plate or strip tube.
• 6
  Dilute the atgRNA inserts (ordered as eBlocks) to ~3ng/μL in UltraPure water. Add 1μL per well to the distributed Gibson master mix.

  • Critical Step: We recommend including a negative control reaction containing digested vector but no insert
• 7
  Use a PCR thermal cycler to incubate the reactions at 50°C for 1 hour,

  • Pause point: Gibson assembly reactions can be stored at −20°C for at least 1 week.

Transformation and purification of assembled products

• 8Transform completed Gibson assembly reactions into a chemically competent E.coli cell line according to the manufacturer’s protocol. Thaw a 25uL aliquot of cells on ice until liquid, add 2μL of the cloning reactions and incubate on ice for 10 minutes. Heat shock the mixtures in a water bath set to 42°C for 45 seconds and immediately place back on ice for 5 minutes. Add 35μL SOC medium and drip-streak 8μL of this mixture per construct on ampicillin-containing agar plates. Incubate the plates overnight at 37°C.
• 9The next day, inspect the plate for colonies. There should be >20 white colonies present on the plate, with minimal colonies on the negative control.

  Critical step: The acceptor plasmid contains an RFP which is removed by successful insertion of the atgRNAs. Red colonies, therefore, indicate undigested backbone and should be avoided.

  Troubleshooting

• 10Using a sterile pipette tip, pick 3–4 white colonies per construct and inoculate each into 1 mL of ampicillin-containing Terrific Broth in a deep well 96-well plate. Seal the plate with gas-permeable aeroseal and incubate it at 37°C for 12–16 hours with shaking at >250rpm.
• 1112–16 hours after inoculation, purify the plasmids using an endotoxin-free miniprep kit, following the manufacturer’s protocol. We prefer using the Qiagen endotoxin free miniprep buffer in conjunction with EconoSpin 96-well miniprep plates.

  Critical step: Use a miniprep kit that ensures endotoxin-free plasmid preparations. These plasmids will be used to transfect mammalian cell lines. High doses of endotoxins can be stressful to mammalian cells and should be avoided.

Validation of insertion by sequencing.

• 12Sanger sequence the purified plasmids using the hU6-F or LKO.1–5′ primer (see Table 1). Verify insertion of the atgRNA between the U6 promoter and Ori sequence. Tools such as Snapgene or Geneious Prime can be used to compare the Sanger output to the expected cloning product.

  Pause point: Purified plasmids can be stored at ‒20°°C for at least 1 year.

  Troubleshooting

Cloning nicking guides (optional) (Timing = 4d)

PCR amplification

• 13Set up a PCR using a forward primer containing the restriction site and nicking guide spacer sequence as an overhang and a reverse primer that amplifies the Cas9 scaffold sequence. As a template, use any plasmid that contains the scaffold sequence for spCas9 (e.g. ACTB nicking guide pDY0209 (Addgene ID: 179109). Use a different forward primer for each nicking guide to be cloned; the reverse primer is universal (see Table 1 for primer sequences).

Component	Amount per reaction (μL)	Final concentration
NebNext PCR Master Mix (2x)	12.5	1x
Nicking guide PCR F (50μM)	0.1	0.2μM
Nicking guide PCR R (universal) (50μM)	0.1	0.2μM
Template (10ng/μL)	1
UltraPure water	11.3
Total volume	25

• 14Incubate the PCRs in a thermal cycler using the following conditions:

Cycle number	Denature	Anneal	Extend
1	98 °C, 30 s
2 – 26	98 °C, 10 s	63 °C, 30 s	72 °C, 10 s
27			72 °C, 2 min

• 15Check a subset of the PCRs by diluting 5–10μL of PCR product in 20μL of water, then running 20μL per lane on a 2% e-gel and confirm correct amplification at ~125bp

PCR purification.

• 16(If the PCRs produced a single, bright band in Step 15, the user has the option of proceeding to the next step without purification of the PCR or to purify the PCRs with columns such as Qiagen Qiaquicks (purification will increase the efficiency of the Golden Gate assembly reaction in the next section). If the PCR produced multiple bands, either further optimize the PCRs, or excise the band at ~125bp and purify with, for example, Monarch Gel Purification Columns.

Golden gate assembly and validation.

• 17Set up a master mix using the following recipe, scaling up as appropriate:

Component	Amount per reaction (μL)	Final concentration
NEB T4 DNA ligase Buffer (10x)	0.5	1x
pU6-pegRNA-GG-acceptor (100ng/μL) (Addgene ID: 132777)	0.25	5ng/μL
PCR product from Step 16	1
Eco31l FastDigest	0.25
NEB T4 DNA ligase	0.125
UltraPure water	To 5

!Critical step: Negative controls lacking the insert are not required when performing Golden Gate assemblies

• 18Incubate the reactions in a thermal cycler using the following conditions:

Temperature (°C)	Time (mins)	Cycles
37	5	15
20	5	15
4	hold	1

• 19Transform 2μL of each reaction into Stbl3 bacteria and validate correct insertion as in Steps 8–12.

Cloning custom payload into acceptor plasmid (optional) (Timing = 4d)

CRITICAL: This section should be done when a payload other than EGFP is desired.

Restriction digest of plasmid backbone.

• 20Digest the payload acceptor plasmid (PASTE payload acceptor with attP site pDY2087 (Addgene ID: 219859)) with the restriction enzymes XbaI and BshTI, which will cut out the GFP payload and allow replacement with the user’s desired payload. Set up the following reaction as a master mix, scaling up depending on the required amount of purified digested plasmid (we typically expect a 30–40% recovery rate):

Component	Amount per reaction (μL)	Final concentration
FastDigest Buffer (10x)	2	1x
Plasmid pDY2087 (2μg/μL)	1.5	100ng/μL
FastDigest XbaI	1
FastDigest BshTI	1
UltraPure water	To 20

• 21Distribute the master mix into 20μL aliquots and incubate at 37°C for 1 hour.
• 22Upon completion of the digest, pool the reactions from Step 21 and run them on a 2% e-gel (~500ng of digest per lane). Successful digestion will produce 2 bands at ~4300bp and ~680bp.
• 23Excise the ~4300bp band from the gel and purify, following the manufacturer’s protocol. We prefer to use the NEB Monarch DNA Gel Extraction kit and modify the protocol to maximize yield by running the dissolved gel through the column twice and eluting twice with ~12μL of water pre-warmed to 65°C. Quantify the purified product using a NanoDrop UV spectrophotometer. Proceed only if the 260/280 absorbance ratio is >1.8. Pause point: purified digests can be stored at −20°C for at least 2 weeks.

PCR amplification of insert.

• 24Amplify the desired insert with PCR primers containing overhangs that are complementary to 5′ and 3′ sequences of the payload plasmid, which will be used for Gibson assembly (see Table 1). Set up a master mix with the following components on ice:

Component	Amount per reaction (μL)	Final concentration
NebNext PCR Master Mix (2x)	12.5	1x
Primer Payload-Gibson-F (Table 1) (50μM)	0.1	0.2μM
Primer Payload-Gibson-R (Table 1) (50μM)	0.1	0.2μM
Template (10ng/μL)	1
UltraPure water	11.3
Total volume	25

Critical step: If amplifying using a genomic DNA template, increase the template amount up to 100ng per reaction.

Critical step: The primers used here contain a 5′ constant region, which is homologous to the payload destination vector used for Gibson assembly, and a primer binding sequence complementary to the insert to be subcloned into PASTE payload acceptor with attP site pDY2087 (Addgene ID: 219859) in N’s. The user should replace the ‘N’ sequences to be complementary to the payload for insertion and utilize these primers for PCR. Note that the R primer will insert a 3′ STOP codon to the payload (‘TCA’ - underlined in the table). If the user does not wish for this STOP codon to be present, simply exclude the TCA sequence.

• 25Incubate the reactions with the following conditions:

Cycle number	Denature	Anneal	Extend
1	98 °C, 30 s
2 – 26	98 °C, 10 s	50–72 °C, 30 s	72 °C, 30 s/kb
27			72 °C, 2 min

PCR Purification.

• 26When the PCRs are complete, run 2μL on a 2% e-gel to check for amplification at the correct size. If the PCRs produced a single, bright band, purify using Qiagen QiaQuick PCR purification kit, following the manufacturer’s protocol. If the PCRs produced multiple amplicons, either optimize the PCR or excise bands running at the correct size and purify using NEB Monarch DNA Gel Extraction Kit, following the manufacturer’s protocol. Quantify the purified product using NanoDrop UV spectrophotometer.

Gibson assembly and verification.

• 27Set up a master mix for the Gibson reactions on ice. Include negative control reactions with water replacing the insert. Adjust the volume of the insert such that it is 2–3 molar excess if >200bp and 5-fold excess if <200bp.

Component	Amount per reaction	Final concentration
Gibson HiFi Master Mix (2x)	5μL	1x
Digested plasmid backbone from Step 23	25ng
Insert from Step 26	variable
UltraPure water	To 10μL

• 28Incubate the reactions at 50°C for 30 minutes. Extending incubation to 60 minutes or more increases assembly efficiency but may also result in increased background colonies containing empty backbone.

  Pause point. Gibson reactions can be stored at −20°C for at least 1 week.

• 29Transform Gibson reactions into chemically competent E. coli cells, then pick, miniprep and sequence as in Steps 8–12, using a primer such as pCI-rev (Table 1) for Sanger sequencing.

  Carefully inspect the sequencing results to confirm correct insertion.

  Pause point: Purified plasmids can be stored at −20°C for at least 1 year.

Transfection of atgRNA combinations (Timing = 3d)

Critical: Screening of atgRNA combinations (effectively, the attB prime editing rate) can be a useful initial step for generating efficient overall PASTE editing at a new target. We recommend screening atgRNAs without payload to simplify both the transfection and sequencing/analysis steps, then transfecting with payload in a second set of experiments. This approach allows easier readout of large numbers of combinations, especially if ddPCR would be used as a primary readout for integration.

Critical: While we have found Lipofectamine 3000 to be a generally good transfection reagent, other reagents (e.g. Trans-IT, PEI) may be more suitable for certain applications, cell lines and cargo sizes.

HEK293FT maintenance

• 30. Culture HEK293FT cells in D10 medium in a humidified incubator at 37°C and 5% CO₂ according to the supplier’s protocol.
• 31To passage HEK293FT cells, aspirate the D10 medium and gently rinse the cells by adding 5mL of PBS to the side of the T75 flask without dislodging the cells. Swirl the PBS and aspirate. Add 2.5mL of TrypLE to the side of the T75 flask, swirl the flask to distribute the TrypLE equally, and place the flask in the incubator. Wait for the cells to detach (approximately 3–4 minutes typically) and return the cells to the laminar flow hood.
• 32Promptly add 7.5mL pre-warmed D10 to the detached cells, tilt the flask and rinse the surface of the flask a couple of times, then collect the cells into a 10mL falcon tube.

  Critical step. We recommend passaging HEK293FT cells every 3–4 days at a ratio of about 1:4 or 1:6. Do not let the cells reach >70% confluency.

Seeding HEK293FT cells for transfection.

• 33Ensure the dissociated cells are well resuspended, count them using a cell counter, and calculate the concentration of cells. Dilute the cells in pre-warmed D10 to a density of 150,000 cells / mL. Adjust the total volume depending on how many plates you wish to seed; we recommend about 14mL per 96-well plate.
• 34Using a reservoir and p200 multichannel pipette, seed 100μL of the diluted cell suspension into each well (which equals 15,000 cells/well) of a 96-well plate and place them in the incubator.

  Transfect the cells about 16–20 hours later.

  CRIITCAL STEP: Cells at 70–80% confluency are at an ideal density of transfection.

  Transfecting the cells at the recommended density is required for maximum transfection efficiency. Under confluency will result in toxicity of the lipofectamine to the cells; over confluency will greatly reduce transfection efficiency. Additionally, the cells should be evenly spaced and not clumped together.

  CRTITICAL STEP: After seeding, ensure the cells are evenly distributed by moving the plates in a figure-of-eight motion

Transfection.

• 35We recommend starting by normalizing all the atgRNA plasmids to the same concentration of 10ng/μL in UltraPure water.
• 36To screen combinations of atgRNAs, lay out all atgRNAs to be paired (i.e. TwinPE atgRNAs targeting sense and antisense strand, or atgRNA/nicking sgRNA combinations) along a row and column of a 96-well plate, respectively. Then, using a P20 multichannel pipette and on a new 96-well plate, multi-channel pipette the rows and columns into each other to create a grid containing an all-by-all 1:1 cross of sense and antisense-targeting atgRNAs, with each well containing a unique combination (Fig. 4b).
• 37Prepare the following mix and scale up accordingly. We recommend making a master mix containing the Opti-MEM, PASTE machinery (e.g., pDY0216) or negative control (pDY0191) and p3000 and, using a reservoir and multi-channel pipette, distribute this in the same layout as the atgRNA combinations from the previous step). Then, multichannel pipette the atgRNA combinations from Step 36) into the distributed master mix. Perform each transfection in at least technical duplicates (ideally, triplicates).

Component	Amount per well
PASTE plasmid (e.g. pDY0216) or pDY0191	100ng
atgRNA cross	20ng
p3000	0.66μL
Opti-MEM	To 5μL

• 38Prepare a master mix containing the lipofectamine and scale up accordingly:

Component	Amount per well
Lipofectamine 3000	0.3μL
Opti-MEM	To 5μL

• 39Mix the Opti-MEM/Lipofectamine 3000 from Step 38 and Opti-MEM/DNA from Step 37 in 1:1 ratio and mix by pipetting. Incubate this mixture for 15 minutes.

  Critical step: During and after this 15 minute incubation, do not mix or vortex the samples as this can disrupt the lipofectamine-DNA complexes that have formed and impair transfection efficiency.

• 40Gently add the transfection mixture to cells (10μL per well) from Step 34. Swirl to distribute and return the cells to the incubator. Harvest the cells for analysis 3 days later.

  ?Troubleshooting

• 41Approximately 72 hours post-transfection, thaw an appropriate amount of Quick Extract buffer on ice. 50μL are needed per well at the 96-well scale.
• 42Remove media from all transfected wells using a multi-channel pipette and discard.
• 43Wash wells with 50mL of room-temperature PBS.
• 44Multi-channel pipette 50μL of Quick Extract buffer onto wells and incubate at room temperature (15–25°C or 68–74 °F) for 3 minutes.
• 45Following room temperature incubation, vigorously scrape wells with a multichannel pipette to dislodge cells, pipette up and down 2–3 times, and transfer the full volume to a thermocycler compatible plate, such as an Axygen 96-well, or set of strip tubes.

  Critical step. A 3-minute incubation is sufficient for HEK293T or similar cells, but a longer incubation in Quick Extract buffer (5–7 minutes) may be necessary for more adherent cell lines.

  Wells should be checked under a microscope after scraping to make sure most cells have lifted.

• 46Incubate in a thermocycler using the following settings.

Temperature (°C)	Time	Cycles
65	15 minutes	1
68	15 minutes	1
98	10 minutes	1
4	Hold	1

Pause point. Genomic DNA isolated from Quick Extract can be stored at −20°C for several months.

Reading out attB insertion by 2-primer NGS (Timing = 2d)

• 47Prepare the master mix for round 1 PCR according to the following table. Scale up according to the number of samples to be amplified.

Component	Amount per reaction (μL)	Final concentration
NebNext PCR Master Mix (2x)	2.5	1x
Primer pool F (10μM each; Table 1)	0.25	0.5μM
)Primer R (100μM; Table 1)	0.025	0.5μM
UltraPure water	To 4.5

Critical step: For Illumina sequencing, we recommend making a pool of forward primers containing at least 4 staggers (see Table 1) to maintain flow cell diversity.

• 48Using a multi-channel pipette, distribute 4.5μL of the NGS master mix per well across a 96-well plate.
• 49Pipette 0.5μL of the genomic DNA from step 46 per well into the plate. Mix well by vortexing and briefly spin the plate down.
• 50Run the first round of NGS PCRs with the following settings in a thermocycler:

Cycle number	Denature	Anneal	Extend
1	98 °C, 30 s
2 – 13 (ACTB) or 23 (NOLC1)	98 °C, 10 s	65 °C, 30 s	72 °C, 30 s
14 (ACTB) or 24 (NOLC1)			72 °C, 5 min

!Critical step. Note that we use different cycling parameters depending on if we are assaying the NOLC1 or ACTB locus.

• 51Prepare the master mix for round 2 PCR according to the following table. Scale up according to the number of samples to be amplified:

Component	Amount per reaction (μL)	Final concentration
NebNext PCR Master Mix (2x)	2.5	1x
UltraPure water	To 4μL

• 52Pipette 4μL of mastermix per well across a new 96-well Axygen plate.
• 53Using a multichannel pipette, add 0.5μL of premixed F and R barcoded primers (see Table 2; 1.25μM each) to each well, from the plate prepared as described in the Reagent Setup and Fig. 4b.

  Critical step. It is imperative that each sample receives a different combination of the forward and reverse primers to facilitate demultiplexing after sequencing. Preparation of the round 2 barcodes as in Fig. 4b ensures that each well of 96-well plate is uniquely indexed.

• 54Briefly spin down the plate from Step 50 and directly pipette 0.5μL of the round 1 PCR into the plate from Step 53 using a multichannel pipette for the round 2 NGS. Vortex the plate to mix and briefly spin down.
• 55Run the 2nd round of NGS according to the following parameters on a thermocycler:

Cycle number	Denature	Anneal	Extend
1	98 °C, 30 s
2 – 13 (NOLC1) or 2–23 (ACTB)	98 °C, 10 s	65 °C, 30 s	72 °C, 30 s
14 (NOLC1) or 24 (ACTB)			72 °C, 5 min

• 56After the second round of NGS finishes, pool wells by taking 2μL from each well and combining in a reservoir. Since all wells are uniquely barcoded, there is no need to change pipette tips between samples.

  Critical step. Only pool samples that are the same round 1 PCR amplicon. For example, NOLC1 amplicons should not be pooled with ACTB amplicons.

• 57Add water to the reservoir equal to the volume of pooled samples and collect the diluted pool into an eppendorf tube. Rinse the reservoir several times with the pooled samples.
• 58Run the amplicons on 3–4 lanes of a 2% E-gel (10μL per lane).

  Critical step. To ensure that the size of the amplicon can be measured accurately, avoid overloading the lanes of the gel. It may be helpful to run several dilutions of the pooled PCR amplicons in water.

• 59Verify that the PCR amplicon matches the predicted length and excise the correct band from the gel.

  Critical step. The user must know precisely the expected size of the amplicon after round 2 PCR and excise only this band from the gel. This should be calculated as the size of the round 1 amplicon plus the additional overhangs that were added during round 1 and round 2 PCR (69 bp). Excision of other bands could greatly impact the quality of subsequent Illumina sequencing.

  ?Troubleshooting

• 60Purify the PCR products using an NEB Monarch DNA Gel Extraction Kit, following the manufacturer’s protocol.
• 61Sequence the gel-extracted samples on an Illumina MiSeq system according to the user manual with 220 cycles of read 1, 8 cycles of index 1, 8 cycles of index 2, and 80 cycles of read 2. We recommend aiming for approximately 10,000 reads per sample (~960,000 reads for a full 96-well plate).

Further optimization of atgRNAs (optional)

• 62For TwinPE, after identifying pairs of atgRNA spacers (or combination of an atgRNA and nicking sgRNA,) that are effective for attB insertion, the user can further optimise their atgRNAs by varying the length of the PBS while keeping the spacer sequences constant. We have found that this can double or triple attB insertion efficiency so, if optimal insertion is desired, we highly recommend performing this second round of atgRNA optimisation. To do this, the user should vary the length of the PBS between 8–15 nt (Fig. 3) and clone them as in Steps 1–12. If using twinPE (i.e. pairs of atgRNAs), varying the PBS in both the sense and antisense strand targeting atgRNA is recommended, followed by an all-by-all cross during transfection as in Steps 31–47. Read out effective attB insertion by NGS as in Steps 48–62. Another optimization that can be considered is adjusting the overlap length of the attB overhangs on the paired atgRNAs. Typically, we start with full overlap, but varying the overlap length has in some cases resulted in higher efficiency editing. It is important to note that when adjusting the overlap, the fidelity of the insert must be analyzed, as insert truncations and indels can occur with any designs.

Transfection with payload plasmid (Timing = 3d)

CRITICAL: At this stage, the user should have identified an optimal pair of atgRNAs for insertion of attB sites at the desired genomic locus. Next, we transfect cells with these atgRNAs alongside a payload plasmid (either the provided pDY0181 or the custom payload plasmid cloned in Steps 20–29).

• 63Repeat Steps 30–34 to culture HEK293FT and seed them in 96-well plate format.
• 64Prepare the following transfection mixture and scale up accordingly. We recommend the user to optimize the amount of payload plasmid used per well, particularly if using a custom payload plasmid. For pDY0181, we find this to be 180ng per well, but this may vary between users and plasmids.

Component	Amount per well
PASTE Plasmid (e.g. pDY0216) or pDY0191 (negative control)	100ng
atgRNA F	20ng
atgRNA R	20ng
Payload plasmid (e.g. pDY0181)	180ng
p3000	0.66μL
Opti-MEM	To 5μL

• 65Repeat Steps 37–46.

3-primer NGS assay (Timing = 2d)

• 66Prepare the 3-primer round 1 PCR master mix containing PCR mix, a pool of forward primers containing staggers, and handles for a second round of PCR to add Illumina adapters, reverse primer and the template DNA to be amplified. Use the following recipe for master mix formulation and scale up accordingly. We recommend making a master mix, distributing the reactions using a multichannel pipette, and using a P10 multichannel pipette to add the genomic DNA.

Component	Amount per reaction (μL)	Final concentration
NebNext PCR Master Mix (2x)	2.5	1x
Primer pool F (10μM each; see Table 1)	0.25	0.5μM
Primer R 1 (100μM; see Table 1)	0.025	0.5μM
pDY0181_GFP_insert_NGS_R (PD6893) (100μM; see Table 1)	0.025	0.5μM
UltraPure water	To 4.5μL

Critical step: For 3-primer assays, it is critical that the 2 reverse primers are of equal concentration to avoid PCR bias towards one of the amplicons.

• 67Repeat Steps 48–50, using template DNA from Step 65, and then perform the round 2 PCR, pool, purify and sequence on an Illumina MiSeq system as in Steps 51–55.

Analysis of NGS (Timing = 1d)

Critical: The following steps assume that the user can run CRISPResso2 (https://github.com/pinellolab/CRISPResso2) which can be installed via Bioconda, Docker or local installation.

• 68Download demultiplexed fastq or fastq.gz files to a local folder. If multiple amplicons have been used in the experiment, it is advised to separate the fastq files by amplicon into different folders.
• 69Set up analysis files. To run CRISPResso2 in batch HDR mode, first prepare the parameter file as follows. Prepare a spreadsheet file containing the column headings ‘name’ and ‘fastq_r1’. List the filenames to be analyzed, separated by row, in the column ‘fastq_r1.’ Assign a unique label to each sample in the ‘name’ column. Save the file as a tab-delimited .txt file as ‘batch_file.txt’ or similar. Do this for each directory containing fastq files and save the file within the directory.
• 70Analyze. From the command line, change directory to a folder containing the fastq files to be analyzed and batch parameter file from Step 69. Run CRISPResso2 in batch mode using the command CRISPRessoBatch followed by the parameters in Table 4.

Table 4.

Parameters for CRISPRessoBatch

Parameter	Calls	Purpose
--batch_settings	e.g. batch_file.txt	Specifies the fastq files to be analysed
-a	[insert unedited sequence]	Specifies the reference amplicon sequence
-e	[insert edited sequence]	Specifies the expected edited amplicon sequence
-g	sgRNA spacer sequence
-q	30	Minimum q-score of a read for analysis, all below are discarded
-qwc		Specify where in the amplicons is the editing window expected
--discard_indel_reads	TRUE	Discards reads containing indels, but counts the number of times this occurs. This is a convenient way to calculate the occurrence of indels downstream. However the user can omit this parameter (the default is FALSE) if they desire more detailed analysis of the types of indels occurring.

Critical step: While we find these parameters to be sufficient for analysing our experiments, additional parameters are available to the user, more details for which can be found at https://github.com/pinellolab/CRISPResso2. An example command could be (replace italicised text with actual sequences): > CRISPRessoBatch --batch_settings batch_file.txt -a unedited sequence -e edited sequence -g GCGTATTGCCTGGAGGATGG -q 30 -- discard_indel_reads -qwc 35–55

• 71Once CRISPResso has completed, open the file named ‘CRISPRessoBatch_quantification_of_editing_frequency.txt’ found in the folder named ‘CRISPRessoBatch_on_batch_file.’
• 72For each sample, editing efficiency can be calculated by dividing the HDR value found in the ‘Reads_aligned’ column by the value found in ‘Reads_aligned_all_amplicons.’ To calculate the rate of indel formation, divide the values in the ‘Discarded’ column (if the parameter -- discard_indel_reads was used) by the values in the column ‘Reads_aligned_all_amplicons.’ 73) Repeat Steps 73–77 for all amplicons in the experiment to be analyzed.

  ?Troubleshooting

Assessing payload integration by ddPCR (Timing = 1–2d)

Critical: genomic DNA to be quantified by ddPCR should be first extracted with QE buffer (Steps 42–47) then further purified by magnetic beads, as described in the following steps.

Preparation of magnetic beads.

• 73Vortex the magnetic bead solution thoroughly to ensure that the beads are evenly distributed and wait for at least 30 minutes for the beads to come to room temperature. Briefly vortex the bead solution again before use.

Purification of genomic DNA using magnetic beads.

• 74Using a multi-channel pipette and reservoir, add beads to 90% of the QE sample (v/v). For example, if the QE sample is 50μL, add 45μL of room-temperature magnetic beads.
• 75Mix gently by pipetting up and down several times, then incubate the sample/bead mixture at room temperature for 5–10 minutes, allowing the genomic DNA to bind to the magnetic beads.
• 76To wash the beads, place the plate on the magnetic stand to pull the beads to the side of the wells. Incubate on the magnet for at least 5 minutes.
• 77Use a multi-channel pipette to remove and discard the supernatant and wash the beads by adding 200μL of 70% ethanol to the wells, then carefully pipette out the ethanol.

  Critical step. Avoid resuspending the beads with ethanol by pipetting on the opposite side of the well.

• 78Repeat the wash step for a total of two washes, then remove the supernatant and air-dry the beads for 5 minutes.
• 79To elute the DNA, remove the plate from the magnetic stand and add 35μL of TE or Ultrapure water to the beads. At this stage, the solution can be mixed by pipetting up and down or by sealing the plate and vortexing.
• 80Incubate the mixture at room temperature for at least 5 minutes to elute the genomic DNA from the beads. CRITICAL STEP: Longer incubations (up to 30 minutes) can improve yields.
• 81To transfer the eluted DNA, place the 96-well plate with mixed beads and DNA back on the magnet. Incubate the plate for 5 minutes at room temperature. Transfer the supernatant containing the eluted gDNA to a new plate and discard the beads.

  Critical step. It is important to avoid carryover of beads into the purified DNA, which can interfere with downstream applications (especially ddPCR). As with washing, pipetting from the opposite site of the well to the beads will reduce the amount of bead contamination in your eluted DNA. Additionally, slow pipetting with a smaller volume pipette tip (such as a 2–20μL) can be used to reduce contamination. If needed, repeat this step with the eluted gDNA to remove any carry over beads.

• 82Using a Nanodrop UV spectrophotometer, quantify a few wells of the bead-purified genomic DNA samples.

Preparation of droplet PCR mix.

• 83Set up the following master mix containing ddPCR Supermix for Probes, an F and R PCR primer, a fluorescently labelled probe complementary to the user’s target amplicon, a fluorescently labelled reference probe and restriction enzyme(s) that prevent background amplification of the plasmid that was transfected into the cells. For the provided NOLC1 assays, use the enzyme Eco47III. Scale up the following master mix accordingly based on the number of samples.

Component	Amount per reaction (μL)	Final concentration
ddPCR Supermix for Probes (2x)	12	1x
Primer F (100μM; see Table 1)	0.1	0.4μM
Primer R (100μM; see Table 1)	0.1	0.4μM
FAM probe (100μM; see Table 1)	0.05	0.2μM
ddPCR HEX reference probe (20x)	1.2	1x
FastDigest restriction enzyme 1	0.1
FastDigest restriction enzyme 2 (optional)	0.1
UltraPure water	To 24μL

Critical step: ddPCR probes should contain a 5′ fluorescent probe (e.g. FAM) and a compatible 3′ fluorescence quencher (e.g. Iowa Black from IDT). We include an internal ZEN quencher from IDT in addition. We recommend probes not to have a 5′ G as this can interfere with fluorescence even after hydrolysis. Ensure that the fluorophores used in the target and reference probes (here, FAM and HEX respectively) are compatible. For further design recommendations, please refer to the Bio-Rad ddPCR documentation.

Critical step: The choice of restriction enzyme(s) is critical for successful ddPCR experiments. As the reverse PCR primer will land in the payload plasmid either inserted into the genome or not, the chances of non-specific PCR amplification off the payload is relatively high. We recommend choosing 1 or 2 restriction enzymes that cut the payload plasmid immediately 5′ of the attP sequence to minimise non-specific PCR amplification. For assays using pDY0181 (Addgene ID: 179115), we recommend using the enzyme Eco47III.

• 84Using a multi-channel pipette, distribute the master mix into Bio-Rad ddPCR 96 well plates. The volume to distribute depends on the volume of gDNA which will be added. The total volume per reaction is 24μl.

  Critical step. If using a QX200 AutoDG for droplet generation, the master mix must be pipetted into Bio-Rad branded ddPCR plates as others are incompatible with the automated droplet generator. If using the manual QX200 Droplet Generator, then any brand of PCR plate is sufficient here.

• 85Add 1–2μL of bead-purified genomic DNA from Step 83 to each well, aiming for between 20 and 120ng of genomic DNA per well. It is not important that each well contains the same amount of input genomic DNA, just that it will fall roughly between 20 and 120ng.

Droplet generation.

• 86Using a Bio-Rad QX200 Droplet Generator or QX200 AutoDG system, take 20μL of the PCR mixture from Step 86 and generate droplets, following the manufacturer’s protocol.

  Critical step. The droplets must be distributed into a Bio-Rad branded ddPCR plate.

• 87After droplet generation is complete, visually inspect the samples for successful droplet generation in each well. Seal the plate with PCR plate heat seal in a PX1 PCR Plate Sealer, following the manufacturer’s protocol.

  Pause point. Droplets are stable overnight at 4°C.

ddPCR and acquisition.

• 88Place the sealed plate in a PCR thermal cycler and incubate with the following conditions:

Cycle number	Denature	Anneal/extend
1	95 °C, 5 mins
2–41	94 °C, 30 s	63 °C, 60 s
42	98 °C, 5 mins

Critical step. For all steps, set the thermal cycler to a temperature ramp of 2°C/second, which ensures a more uniform thermal transfer to the droplets.

• 89After completion of the PCR, acquire the samples using a QX200 Droplet Reader, according to the manufacturer’s protocol.

  Critical step. In the acquisition settings, ensure that you have selected the correct probe dyes for the target and reference amplicons (in our protocol, it is FAM and HEX, respectively).

Analysis of ddPCR results.

• 90Once the acquisition has completed, threshold the target and reference droplets according to the manufacturer’s protocol. Calculate the editing percentage of each sample as the percentage of target droplet vs reference droplets.

  ?Troubleshooting

Timing

A workflow and timing for PASTE experiments is provided in Fig. 1

Steps 1–19,	Cloning guide RNAs: 4d
Steps 20–29,	Cloning custom payload into acceptor plasmid: optional): 4d
Steps 30–46,	Transfection of atgRNA combinations: 3d
Steps 47–61,	Reading out attB insertion by 2-primer NGS: 2d
Steps 63–65,	Transfection with payload plasmid: 3d
Steps 66–67,	3-primer NGS assay: 2d
Steps 68–73,	Analysis of NGS: 1d
Steps 74–91,	Assessing payload integration by ddPCR: 1–2d
Box 3, Production of in vitro transcribed PASTE mRNA: optional): 1d

Troubleshooting

Troubleshooting advice can be found in Table 5.

Table 5.

Troubleshooting.

Step	Problem	Possible reason	Solution
9)	Low number of colonies	Inefficient cloning	Ensure plasmid backbone is high quality. Use fresh enzymes and buffer.
		Incorrect overhangs	Check overhang designs. The cloning can be simulated in Snapgene to confirm that the choice of enzymes and overhang sequences are compatible.
	High frequency of background (red) colonies	Inefficient digestion of plasmid backbone	Ensure plasmid backbone is high quality. Use fresh restriction enzymes and buffer.
41)	Poor transfection efficiency	Cell quality is suboptimal	Use low passage cells (<15). Ensure cell density at time of transfection is 70–80%. Do not let cells in culture reach >80% confluency during maintenance. Ensure cells are seeded evenly by swirling the plate in a figure-of-eight motion after seeding. Re-optimise transfection conditions using a fluorescent plasmid e.g., pMAXGFP
		Plasmid DNA is poor quality	Re-prep plasmids / RNA. Carefully inspect them using a Nanodrop. Do not use if they are low concentration and/or 260/280 is <1.8. Run them on a gel to check for degradation.
		Transfection reagent has gone bad	Replace lipofectamine 3000.
	Poor cell health post-transfection	Transfection is stressful, endotoxin contamination in plasmids	Re-prep plasmids using an endotoxin-free prep kit. Try lowering the overall amount of DNA/RNA transfected per well. Confirm cells were at correct confluency at time of transfection.
60)	PCR amplification fails	No amplification of target band	Ensure primer sequences and concentration are correct. Use the primers for NOLC1 listed in this protocol as a positive control. Ensure cell lysis was complete. Further purify gDNA extracts with magnetic beads. Try different volumes of input gDNA. Re-optimize PCR cycling temperatures, particularly the primer annealing temperature. Try re-designing the round 1 primers.
78)	No measurable editing in workhorse cells (e.g. HEK293FT)	Suboptimal atgRNAs	Use positive control atgRNAs, payload plasmid and NGS primers provided in this protocol. If no editing is observed for positive controls, re-prep plasmids, confirm transfection was successful with pMAXGFP, double check correct use of atgRNAs, payloads and NGS primers. If the components provided in this protocol work but not custom atgRNAs, try different spacers, try more PBS lengths, ensure the 5′ end of the spacer is a G, check design of NGS primers and/or ddPCR primers and probe
		Genomic locus is not permissive to editing	Some genomic loci are just difficult to edit. Try more spacer sequences or switch to a new locus. Consider whether attB and/or payload insertion at this site could be toxic to cells.
		CRISPResso2 parameters are incorrect	Inspect fastq files for reads corresponding to editing. In the terminal, use the command '> grep -c [SEGMENT OF EDITED SEQUENCE] *.fastq' to search and count for edited reads (note that the fastq files must be unzipped and the search string in capital letters). If grep finds edited reads but CRISPResso does not measure them, use the CRISPResso2 documentation to check the parameters used.
	High rate of indels	Suboptimal nicking guide RNA	Test more nicking sgRNAs, switch to twinPE editing
91)	No positive droplets for target amplicon	PCR failure	Use atgRNAs, payload, primers and probes listed in this protocol as a positive control. Test new primer/probe designs. Optimise cycling conditions.
	High target signal in negative control wells	Non-specific amplification from the payload	Try additional restriction enzymes that cut the payload plasmid 5′ of the attP. Re-design forward primer.
	Low total droplets	Low fraction of droplets contain successful PCR amplification	Ensure the correct amount of gDNA is added to each sample. Check that the correct concentration of primers and probe was used, optimise if necessary.
		Droplet generation failure	Ensure correct dilution of probe master mix and correct total volume. Ensure all wells in a column contain probe mastermix or buffer (not water, or empty).
96)	Gel electrophoresis of the IVT mRNA shows no band, smear, or incorrectly sized band	Transcription failure	Ensure input DNA has a 5′ T7 promoter and 1 μg of high quality DNA was used as the template. Use fresh enzymes.
Box 3	Gel electrophoresis of the IVT mRNA shows no band, smear, or incorrectly sized band	RNAse contamination	Repeat IVT protocol with RNase free technique

Anticipated Results

After screening of guides and optimizing atgRNA parameters, PASTE editing can result in highly efficient and specific programmable gene integration at desired loci in mammalian cells (Fig. 6). Because optimal guide parameters can vary so much between different target loci, it is important to screen enough designs to cover the overall design space and capture sequences with maximal efficiency. With the provided atgRNAs, alongside the provided NGS and ddPCR reagents, the user should expect to obtain at least 10% (NOLC1) or 30% (ACTB) editing in HEK293FT cells, although higher is possible with optimal cell health, transfection conditions and plasmid quality. When working with insertion templates that are longer (ex. >5kb), it is also advisable to increase the molar amount of template delivered to improve insertion efficiency. Analysis of sequencing data allows for interpretation of the overall attB attachment site placement efficiency (two primer), integration rate (three primer junctional assay), and leftover unintegrated beacons, offering an overview of the different editing processes and their efficiencies. By using CRISPResso2, the sequencing data can further be used for understanding beacon fidelity and the purity of the final integration product. As these parameters can vary drastically between different cell types, it is important to measure and understand these metrics. Confirming payload integration with ddPCR readouts using primers that span the integration junction allows for verification of the three-primer sequencing assay data to offer more confidence in the determined integration efficiency. Finally, verification of the fidelity of the entire integration sequence could also be performed via long read sequencing as an orthogonal way to check for issues across the entire integrated sequence. Because of the multiple components of PASTE editing, the efficiency can vary drastically between different delivery methods (plasmids, RNA, viral, or electroporation) and so delivery should be optimized for each application.

Figure 6: Expected outcomes from PASTE editing.

This figure shows outcomes of PASTE editing at several loci, including intermediate stages of the readout. a) attB insertion rates at endogenous NOLC1 using atgRNAs that insert attBs ranging from 38–46 base pairs, measured by 2-primer NGS. Here, the 38bp attB inserts with the highest efficiency. b) PASTE integration at 4 endogenous loci, ACTB, CCR5, CFTR, and NOLC1 with or without the addition of a prime editor. Both perfect and imperfect (indel) editing is shown for each condition. All four loci show robust editing with minimal indel formation recorded by NGS c) A CRISPResso2 output from the NOLC1 editing experiment in a), showing the genomic target (top) and the 4 most common editing outcomes. d) Editing outcomes at the NOLC1 locus by different activity PASTE mutants, assessed by ddPCR. The gate used to threshold the droplets is show in pink in each plot. Grey dots are denoted as PCR-negative, blue (NOLC1) and green (reference) are PCR-positive. Output is rendered using QuantaSoft (BioRad). All data are mean +/− SEM, n = 3.

Key points:

• PASTE (Programmable addition via site-specific targeting elements) combines the specificity, efficiency, and cargo size advantages of site-specific integrases with the programmability of prime editing for precise and efficient integration of large DNA sequences into mammalian genomes.

• PASTE offers improved editing efficiency, purity and reprogrammability compared to previous methods for long insertions into the mammalian genome. Key references:

Data availability Statement:

Sequencing data used in Fig. 6 are deposited at the NCBI Sequence Read Archive database, PRJNA1101023.