Stem Cell

Robust and inducible genome editing via an all-in-one prime editor in human pluripotent stem cells

Time Stamp: December 29, 2024 7:00 PM
Source Node: 842886

Comparison of current PE tools in human pluripotent stem cells

Multiple prime editing tools have been developed to improve prime editing efficiency in mammalian cells. However, determining the optimal PE tool for use in hPSCs remains uncertain. To address this question, we generated a “H2B-turn-on reporter” for real-time monitoring of prime editing outcomes (Fig. 1a). This reporter construct was generated by introducing an H2B-tdTomato cassette into one allele of SOX2 locus in the hESC H1 line, where a “C” deletion in the H2B sequence was present. This “C” deletion caused a frameshift between the tdTomato gene and the SOX2 gene, resulting in the lack of tdTomato fluorescence. Prime editing was applied to reintegrate the “C” nucleotide within H2B, allowing tdTomato to be expressed in-frame with SOX2, thereby activating fluorescence (Fig. 1b). Therefore, the proportion of tdTomato-positive cells reflects the efficiency of prime editing. In most cases, the additional nicking sgRNA significantly enhances prime editing for endogenous loci compared to PE2^1,27. However, this improvement was not observed in the “H2B-turn-on reporter” system, which showed similar editing efficiency with or without the additional nicking sgRNA (Supplementary Fig. 1). Therefore, no nicking sgRNA was included in the comparison of different conditions in this reporter system. Various prime editing conditions were evaluated based on FACS data (Fig. 1d, e). As expected, PE4 demonstrated enhanced editing efficiency compared to the original PE2, resulting in an increase from 12.2% to 18.0% in tdTomato-positive cells. The utilization of either PE2 or PE4 along with P53DD led to further improvements (23.9% with PE2 + P53DD; 30.3% with PE4 + P53DD) in tdTomato-positive cell population. Notably, the combination of PE4 and P53DD exhibited the highest efficacy among the tested conditions, indicating the additive enhancement resulting from concurrent inhibition of the MMR pathway and P53. By substituting the PE2 enzyme with PEmax, editing efficiency was synergistically amplified in conjunction with the aforementioned four conditions (18.4% with PE2max, 19.8% with PE4max, 29.2% with PE2max + P53DD, 38.0% with PE4max + P53DD). Moreover, we explored two versions of engineered pegRNA (epegRNA), each containing distinct 3’RNA motifs to enhance pegRNA stability²⁵. The combination of either epegRNA, PE4max, and P53DD led to further enhancement in PE efficiency, with more than 55% of cells exhibiting tdTomato fluorescence 48 h after transient transfection (Fig. 1c, d). To confirm our observations using the “H2B-turn-on reporter” system, we performed amplicon sequencing (Miseq) to quantify frequencies of prime editing and byproducts. The addition of P53DD (PE4max + P53DD) or substituting the pegRNA with epegRNA (PE4max + tev or PE4max + tmp) increased prime editing efficiencies of PE4max from 8.4% to 18% ~ 19%. Combining PE4max with P53DD and epegRNA further improved the editing efficiency to 24% ~ 27% (Fig. 1f). Meanwhile, the editing specificity, calculated as the ratio of on-target edits to byproducts, was not compromised (Fig. 1g). Normalization of data from the two experiments, as determined by the fold change of tdTomato-positive cells assessed by FACS (Fig. 1h) and editing frequencies assessed by Miseq relative to PE4max (Fig. 1i), showed high consistency, with approximately two-fold increases when PE4max was combined with either P53DD or epegRNA, and around three-fold increases when PE4max was combined with both P53DD and epegRNA.

**Fig. 1: Comparison of different prime editing tools in hPSCs using an “H2B-turn-on reporter” cell line.**

To further validate the observation from the reporter system that PEmax, MLH1dn, and P53DD yielded the most effective prime editing, we conducted different types of editing at endogenous genomic loci in hPSCs. The combination with MLH1dn and P53DD significantly improved 2nt deletion at the HEK3 site to 10.4%, compared with either PEmax (4.2%) or PEmax with MLH1dn (5.7%) (Fig. 1j). In the cases of larger size editing, MLH1dn did not show any increase of editing efficiency when conducting a 30nt deletion (Fig. 1k) or a 34nt “Loxp” insertion (Fig. 1l) at the HEK3 locus, while the addition of P53DD substantially improved PEmax-mediated 30nt deletion from 3.1% to 12.1% and 34nt insertion from 7.9% to 24.3%. Similar results were observed in the case of a 10nt deletion at the SOX2 locus, showing 0.56% with PEmax, 0.8% with PEmax+MLH1dn, and 2.5% with PEmax+MLH1dn + P53DD (Fig. 1m). Since the precise deletion efficiencies at the SOX2 locus were much lower than those at the HEK3 locus, we applied twin-PE, which is expected to show a higher editing ratio for larger size editing using a pair of pegRNAs on a 40nt deletion at the SOX2 locus²⁸. PEmax and PEmax+MLH1d showed 2.8% and 2.4% ratio of precise 40nt deletion, respectively, while PEmax + MLH1d + P53DD dramatically increased the on-target efficiency to 16.7% (Fig. 1n). This improvement is mainly due to the P53DD, as PEmax combined with P53DD alone showed a similar editing efficiency (15.7%). Taken together, these data indicate that the combination of PEmax, MLH1dn, and P53DD was identified as the most effective PE condition in hPSCs, where MLH1dn and P53DD additively improve the efficiency for smaller edits, and the P53DD component can significantly improve the efficiency for larger edits achieved with the PE platform.

Generation of All-in-one PE-Plus prime editor

We next aimed to construct a primer editor that incorporated all three components found to yield the highest prime editing efficiency in hPSCs, including PEmax, MLH1dn and P53DD. Our first step was the optimization of the P53DD to achieve a more efficient enhancement of PE. The P53DD that was previously assessed in our study was derived by mouse and consisted of the first 13 codons of mouse p53, followed by its C-terminal region containing three domains: nuclear localization signal (NLS) region, tetramerization domain (TET) and C-terminal negative regulatory domain (CTD)^29,30. We asked whether a human-derived P53DD would exhibit greater efficiency in human stem cells (Supplementary Fig. 2a). Similar to the construction of the mP53DD plasmid, the human P53DD plasmid (hP53DD) expresses the C-terminal region of human P53, which is essential for the formation of a stable P53/DNA complex³¹. The cloned C-terminal region functions to inhibit P53 by competition for DNA binding. We cloned the full C-terminal region of human P53, consisting of all three domains: the NLS, TET, and CTD. Meanwhile, truncated hP53DD only contained TET and CTD domains, and a variant with only the TET domain, was also cloned for comparisons (Supplementary Fig. 2b). Using the PE reporter cell line as mentioned above, we assessed PE efficiency in the presence of mP53DD, hP53DD and hP53DD truncations (Supplementary Fig. 2c, d). We observed improvement in PE efficiency across all the P53DD plasmids. Specifically, mP53DD and hP53DD yielded the highest editing efficiency (27.5% with mP53DD and 30.8% with hP53DD), whereas the truncated P53DD variants did not perform as effectively as the full-length hP53DD (23.3% with hP53DD-TET-CTD and 18.8% with hP53DD-TET). These results suggested that both mouse and human P53DD improve prime editing efficiency in hPSCs. Furthermore, all three domains within the C-terminal region were found to contribute to the enhancement of PE editing. This can potentially be attributed to the distinct functions of each domain in DNA binding, and the incorporation of all three domains seems to lead to the most effective inhibition of P53. However, a comprehensive understanding of the underlying mechanisms will require further elucidation. The two versions of P53DD were further assessed by targeting more endogenous loci, including the induction of the N370 mutation at the GBA locus, the L858R mutation at the EGFR locus, the G12C mutation at the KRAS locus, as well as a “LoxP” insertion at the HEK3 locus (Supplementary Fig. 2e). The enhancement effect of the two P53DD versions over PE3 was similar according to the desired edits at all the tested targeting sites (Supplementary Fig. 2f), indicating that either one can be chosen for the PE-Plus system.

Next, we incorporated hP53DD and human hMLH1dn into the PEmax prime editor either through direct fusion or by utilizing linkages in between. Two fusion plasmids were generated on top of PEmax-P2A-hMLH1dn (PE4max), with an additional hP53DD directly fused to N-terminus or C-terminus of PEmax (Fig. 2a). The H2B-turn-on reporter assay showed that the two plasmids with direct fusion of hP53DD reduced the prime editing efficiency, indicating a negative impact of protein fusion on the prime editor (6.7% with N-terminal fusion, 10.8% with C-terminal fusion) (Fig. 2b). In contrast, the plasmid with linkages PEmax-P2A-hP53DD-IRES-hMLH1dn increased editing efficiency compared with PE4max, rising from 17.8% to 28.8%. We further tested the plasmid with two P2A linkages (PEmax-P2A-hP53DD-P2A-hMLH1dn). The data showed no difference in the editing efficiency generated by the plasmids with P2A/IRES or P2A/P2A linkages (Supplementary Fig. 3). We name the robust all-in-one PE editor as PE-Plus (PEmax-P2A-hP53DD-IRES-hMLH1dn).

**Fig. 2: Generation of all-in-one prime editor co-expressing PEmax, hMLH1dn, and hP53DD.**

Analysis of genome-wide off-target effects induced by PE-Plus

The PE-Plus edited cells and PEmax edited cells were then enriched by isolating the tdTomato positive cells after frame restoration in the “H2B-turn-on reporter”. The editing outcomes in the tdTomato-positive cell populations were evaluated by Miseq analysis, which showed that 86% of the sorted cells in either condition exhibited perfect “C” insertion. The rate of unwanted modifications, including “reference modified,” “prime-edited modified” “scaffold-incorporated,” and “ambiguous,” was low (3.05% with PEmax and 2.72% with PE-Plus in total). (Supplementary Fig. 4).

To evaluate the genome-wide safety of the PE-Plus prime editor, we performed whole genome sequencing (WGS) to identify unwanted off-target effects in the genome. Both SNV and indel mutations in the genome were identified by comparing them to the unedited parental cells (Fig. 2d). The number of insertions and deletions was not increased upon PE-Plus treatment (36 insertions and 57 deletions in PEmax-edited cells; 35 insertions and 38 deletions in PE-Plus-edited cells). Similarly, no increase in SNVs was observed upon PE-Plus treatment (322 in PEmax-edited cells vs. 296 in PE-Plus-edited cells) (Fig. 2e). Among the SNV mutations, the two prime editors resulted in a similar mutation pattern in the genome, as analyzed by the number and relative proportions of SNV mutation types (Fig. 2f). Consequently, these data suggest that PE-Plus improves prime editing efficiency without compromising genome-wide safety.

Generation iPE-Plus hPSC cells

To achieve more efficient and versatile prime editing in hPSCs, we generated a prime editing platform for the doxycycline-induced expression of PE-Plus. As this inducible-PE platform did not require transfection, editing efficiency would not be affected by plasmid size or cytotoxicity upon plasmid delivery. The iPE-Plus hPSCs were engineered by TALEN-mediated genome editing targeting the safe harbor locus AAVS1 (Supplementary Fig. 5a, b). One donor plasmid expressing PE-Plus under the control of TRE promoter (TRE-PE-Plus-Hygro) and another donor plasmid expressing M2rtTA under the CAG promoter (CAG-M2rtTA-Neo), were electroporated into hPSCs along with a pair of AAVS1 TALEN vectors (Fig. 3a). To achieve inducible mutations in the genome, iPE-Plus cells were pre-infected with pegRNA and/or nicking sgRNA lentivirus. The intended edits would then be installed into the genome upon doxycycline treatment (Fig. 3a).

**Fig. 3: Inducible prime editing in hPSCs with the iPE-Plus platform.**

To provide a preliminary assessment of iPE-Plus hPSCs, we utilized the “H2B-turn-on reporter” system described above (Fig. 3a, b). Rapid and efficient inducible expression of the prime editor was observed in three selected clones after doxycycline treatment, as demonstrated by qPCR (Supplementary Fig. 5c). The iPE-Plus reporter cells were then infected with either pegRNA or epegRNA lentivirus, followed by doxycycline treatment for different durations to active tdTomato (Fig. 3b). The tdTomato positive cells was checked at various time points following doxycycline addition. In the presence of pegRNA, the population of tdTomato + cells increased to 35.1% at day 2 and further to 69.0% at day 4. The increase then slowed down after day 4, gradually reaching 85.8% at day 8 (Fig. 3c–e). In cells infected with epegRNA lentivirus, tdTomato was switched on more rapidly, with 77.5% at day 2, reaching a plateau after day 4 (Fig. 3c–e). Furthermore, to access induced prime editing in an endogenous gene, we applied this platform with an epegRNA lentivirus to insert a “TGA” stop codon into the SOX2 gene in the heterozygous H1-SOX2-tdTomato reporter line, thereby silencing tdTomato expression (TGA-insertion reporter) (Supplementary Fig. 6a.) FACS data and fluorescence images demonstrated the gradual switch-off of tdTomato over 12 days. Approximately 23.7% of cells became tdTomato negative after 2 days of dox treatment, increasing to 65.1% at day 4, and eventually reaching 90% at day 10 (Supplementary Fig. 6b–d). These results collectively demonstrate inducible and time-dependent prime editing by the iPE-Plus platform, controlled by doxycycline treatment.

To study whether inducible prime editing can also be temporally controlled during hPSC differentiation, we conducted neuroectoderm differentiation using the SOX2-H2B-turn-on reporter that carries the iPE-Plus platform. We generated neuron progenitor cells via dual SMAD inhibition³² in 7 days, and maintained the NPCs for additional 7 days (Fig. 3f). Doxycycline was added to the cells during 7 days of Neuroectoderm induction or was added to NPCs from Day7 to Day 14 (Fig. 3f). Given that SOX2 is highly expressed in the neural progenitor cells (NPCs), the successful editing can be evaluated via tdTomato turn-on during the differentiation. Immunofluorescence staining demonstrated that a large population of cells at day 7 or day 14 of doxycycline treatment co-expressed PAX6 and tdTomato. Over 80% of tdTomato-positive cells were detected upon doxycycline treatment at the two different stages (Fig. 3h, i) measured by FACS. This data showed the H2B editing could be robustly induced in the NPC differentiation process, as well as in hESC-derived NPCs. The editing efficiency in NPCs was similar to that in the undifferentiated hPSCs (Fig. 3d, e).

Inducible disease-related mutations and evaluation

Next, we employed the iPE-Plus platform for the generation of disease-related mutations in hPSCs. In parallel, we created cells with inducible PE (iPE2) and inducible PEmax (iPEmax) (Supplementary Fig. 7). Two single clones were selected for each of the inducible cell lines and infected cells with epegRNA and sgRNA lentivirus designed to induce disease-related mutations: the GBA N370S mutation associated with Parkinson’s disease (PD)³³ and EGFR L858R mutation linked to cancer³⁴. To measure editing efficiency at various time points after doxycycline treatment, we utilized droplet digital PCR (ddPCR) to quantify mutation rates. In ddPCR, we designed a FAM-labeled probe to bind to the mutation sequence, while a HEX-labeled probe was designed to bind to a non-targeted sequence within the same amplicon (Fig. 4a). The ratio of FAM positive to HEX positive events provides the mutation rate. In the absence of doxycycline, the mutation rates were undetectable, and all three inducible PE cell lines exhibited time-dependent induction of mutations as incubation time increased (Fig. 4b, c). Consistent with data generated by nucleofection, iPE-Plus demonstrated the highest editing efficiency compared to iPE and iPEmax cells. After 7 days of doxycycline induction, iPEmax and iPE cells showed a GBA N370S mutation rate of 21.7% and 16.8%, respectively (Fig. 4b). and EGFR L858R mutation rates of 28.2% and 18.2%, respectively (Fig. 4c). In contrast, the desired mutation rates induced by iPE-Plus cells reached close to 50% for both GBA N370S and EGFR L858R mutations (Fig. 4b, c).

**Fig. 4: Inducible installation of disease-related mutations in hPSCs.**

Editing frequencies from cells with 7 days of doxycycline induction were also measured by Miseq, and the editing efficiencies closely matched the results from ddPCR (Fig. 4d–f). Meanwhile, the proportion of unwanted byproducts during prime editing was observed to slightly increase along with the intended edits. Nevertheless, the edit/byproduct ratio was similar between the three cell lines when inducing GBA N370S mutation (7.26 with iPE, 8.4 with iPEmax and 7.7 with iPE-Plus) and even higher for iPE-Plus when inducing the EGFR L858R mutation (3.4 with iPE, 4.8 with iPEmax and 6.1 with iPE-Plus) (Fig. 4g). The improvement of prime editing with iPE-Plus over PEmax was also observed when introducing another PD-related G2019S mutation at the LRRK2 locus³⁵, increasing mutation rate from 9.9% to 18.5% (Fig. 4h), and a “Loxp” insertion at the HEK3 locus, increasing from 42.8% to 56.7% (Fig. 4i), without sacrificing editing purity (Fig. 4j). We further applied iPE-Plus to induce KRAS mutations, the most common gene mutations related to cancer^36,37. After 7 days of doxycycline induction, over 4% of G12C mutations and around 12% of G13A mutations were generated (Supplementary Fig. 8).

To further characterize induced prime editing in hPSCs, single-cell-derived clones were genotyped by Sanger sequencing from cells treated with doxycycline for 7 days. For GBA N370S mutation induction, 6 out of 30 clones from iPE cells were edited, all of which were heterozygous. iPEmax cells yielded 4 out of 40 clones with heterozygous edits and 7 out of 40 clones with homozygous edits. Among the 38 iPE-Plus clones, 2 were heterozygous, and 17 clones were homozygous (Fig. 5a–c). Similarly, for EGFR L858R mutation induction, 10 out of 40 cells had desired edits by iPE, all of which were heterozygous. iPEmax clones comprised 12 out of 40 edited clones, with 9 heterozygous and 3 homozygous. Among the 38 iPE-Plus clones, 13 were heterozygous, and 13 were homozygous (Fig. 5d–f). These data indicate that the proportion of edited clones increased from iPEmax cells compared with iPE cells, and further increased when using iPE-Plus cells, with a notable improvement in the efficiency of homozygous edits.

**Fig. 5: Generation of single-cell clones carrying disease-related mutations using iPE, iPEmax and iPE-Plus platforms.**

Generation of multiplex disease-mutations in hPSCs

Next, we investigated whether the iPE-Plus platform could be applied to achieve efficient one-step multiplexed prime editing. Our aim was to create a clinically relevant disease model in hPSCs by introducing two mutations at the EGFR locus: L858R and T790M mutations. The L858R mutation at the EGFR locus is one of the most common mutations in non-small-cell lung cancer (NSCLC) patients and can be targeted by EGFR tyrosine kinase inhibitors (EGFR-TKIs)³⁸. However, the presence of the secondary T790M mutation leads to resistance to earlier generations of EGFR-TKIs, reducing median survival to less than 2 years after acquiring this mutation^39,40. To induce the dual mutations simultaneously, we generated a lentivirus vector expressing dual pegRNAs and another vector expressing dual sgRNAs (Fig. 6a). Mutation rates for each EGFR mutation at different time points of doxycycline induction were measured by ddPCR and revealed a time-dependent accumulation of the two mutations (Fig. 6b). After 7 days of doxycycline induction, the mutation rates of L858R and T790M were approximately 20% and 25%, respectively (Fig. 6b, c). Different types of byproducts were observed to increase during the editing process, especially with the T790M mutation. This was also noted in the single-cell clone genotyping, where 6 out of 42 clones contained unwanted sequences, and 5 of them carried mutation near the targeting site for T790M induction (Fig. 6d). However, the higher byproduct rate was not attributed to multiplex PE editing, as the similar results were observed during single T790M mutation induction (Supplementary Fig. 9). Nevertheless, we successfully obtained single clones carrying different mutations with dox treatment, including monoallelic and biallelic single T790M mutations or single L858R mutations. Moreover, we obtained 5 clones with dual mutation, one with heterozygous L858R along with heterozygous T790M and four with heterozygous L858R along with homozygous T790M. These single clones carrying multiple mutations offer a potential disease modeling platform for studying the mechanism of these mutations and evaluating new therapies.

**Fig. 6: One-step installation of multiplex mutations in hPSCs by the iPE-Plus platform.**

SEO Powered Content & PR Distribution. Get Amplified Today.
PlatoData.Network Vertical Generative Ai. Empower Yourself. Access Here.
PlatoAiStream. Web3 Intelligence. Knowledge Amplified. Access Here.
PlatoESG. Carbon, CleanTech, Energy, Environment, Solar, Waste Management. Access Here.
PlatoHealth. Biotech and Clinical Trials Intelligence. Access Here.
Source: https://www.nature.com/articles/s41467-024-55104-1

Latest Intelligence

Apoptotic clearance by stem cells: molecular mechanisms for recognition and phagocytosis of dead cells

December 29, 2024 7:00 pm

Reconstitution of pluripotency from mouse fibroblast through Sall4 overexpression

December 29, 2024 7:00 pm

DDX18 coordinates nucleolus phase separation and nuclear organization to control the pluripotency of human embryonic stem cells

December 29, 2024 7:00 pm

miRNA changes associated with differentiation of human embryonic stem cells into human retinal ganglion cells

December 29, 2024 7:00 pm

Weekly reads: coral stem cells to the rescue, diabetes

December 29, 2024 4:28 am

Single-cell transcriptome analysis of stem cells from human exfoliated deciduous teeth investigating functional heterogeneity in immunomodulation

December 27, 2024 7:00 pm

Histological effects of combined therapy involving scar resection, decellularized scaffolds, and human iPSC-NS/PCs transplantation in chronic complete spinal cord injury

December 27, 2024 7:00 pm

Robust and inducible genome editing via an all-in-one prime editor in human pluripotent stem cells

Comparison of current PE tools in human pluripotent stem cells

Generation of All-in-one PE-Plus prime editor

Analysis of genome-wide off-target effects induced by PE-Plus

Generation iPE-Plus hPSC cells

Inducible disease-related mutations and evaluation

Generation of multiplex disease-mutations in hPSCs

Latest Intelligence

Apoptotic clearance by stem cells: molecular mechanisms for recognition and phagocytosis of dead cells

Reconstitution of pluripotency from mouse fibroblast through Sall4 overexpression

DDX18 coordinates nucleolus phase separation and nuclear organization to control the pluripotency of human embryonic stem cells

miRNA changes associated with differentiation of human embryonic stem cells into human retinal ganglion cells

Weekly reads: coral stem cells to the rescue, diabetes

Single-cell transcriptome analysis of stem cells from human exfoliated deciduous teeth investigating functional heterogeneity in immunomodulation

Histological effects of combined therapy involving scar resection, decellularized scaffolds, and human iPSC-NS/PCs transplantation in chronic complete spinal cord injury

iPSC-derived human sensory neurons reveal a subset of TRPV1 antagonists as anti-pruritic compounds

Curcumin liposomes alleviate senescence of bone marrow mesenchymal stem cells by activating mitophagy

Remote disruption of intestinal homeostasis by Mycobacterium abscessus is detrimental to Drosophila survival