Therapeutic gene correction for Lesch-Nyhan syndrome using CRISPR-mediated base and prime editing

Abstract

Lesch-Nyhan syndrome (LNS) is inherited as an X-linked recessive genetic disorder caused by mutations in hypoxanthine-guanine phosphoribosyl transferase 1 (HPRT1). Patients with LNS show various clinical phenotypes, including hyperuricemia, gout, devastating behavioral abnormality, intellectual disability, and self-harm. Although uric acid overproduction can be modulated with the xanthine oxidase inhibitor allopurinol, there exists no treatment for behavioral and neurological manifestations of LNS. In the current study, CRISPR-mediated base editors (BEs) and prime editors (PEs) were utilized to generate LNS-associated disease models and correct the disease models for therapeutic approach. Cytosine BEs (CBEs) were used to induce c.430C>T and c.508C>T mutations in HAP1 cells, and then adenine BEs (ABEs) were used to correct these mutations without DNA cleavage. PEs induced a c.333_334ins(A) mutation, identified in a Korean patient with LNS, in HAP1 cells, which was corrected in turn by PEs. Furthermore, improved PEs corrected the same mutation in LNS patient-derived fibroblasts by up to 14% without any unwanted mutations. These results suggest that CRISPR-mediated BEs and PEs would be suggested as a potential therapeutic strategy of this extremely rare, devastating genetic disease.

Graphical abstract

Lesch-Nyhan syndrome (LNS) is a rare genetic disorder caused by a deficiency of HPRT, and there are no fundamental treatment modalities for LNS. Jang et al. demonstrate that CRISPR-mediated base editing and prime editing were utilized to generate LNS-associated disease models and correct the disease models for therapeutic approach.

Introduction

Lesch-Nyhan syndrome (LNS; OMIM: 300322) is a rare X-linked recessive disorder caused by a deficiency in hypoxanthine-guanine phosphoribosyl transferase 1 (HPRT1), which is encoded by the HPRT1 gene at Xq26.^1,2 To date, more than 600 disease-associated mutations have been identified in LNS (http://www.lesch-nyhan.org). The estimated prevalence of LNS is approximately 1 in 380,000 live births.³ HPRT plays a crucial role in the purine salvage system by catalyzing the conversion of hypoxanthine and guanine to their respective 5′ mononucleotides, inosine monophosphate (inosinic acid) and guanine monophosphate (guanylic acid).⁴ HPRT deficiency causes hyperuricemia and a wide spectrum of neurological symptoms such as motor dysfunction, cognitive impairment, behavioral disorder, and self-injurious behavior. Although the causal role of HPRT1 mutations in LNS has been confirmed over the past 50 years, there are no fundamental treatment modalities for LNS.²

CRISPR-Cas-based genome editing tools are highly promising for the advancement of gene therapy, as they are able to correct mutations in endogenous DNA.⁵ Particularly, CRISPR-mediated base editors (BEs) can directly convert targeted base pairs without inducing DNA double-strand breaks. Cytosine BEs (CBEs),^6,7 which comprise catalytically modified Cas9 with cytosine deaminase, can induce C:G to T:A conversion, and adenine BEs (ABEs), which comprise catalytically modified Cas9 with adenine deaminase, can induce A:T to G:C conversion.^8,9 These BEs are widely used for therapeutic applications for liver disease,^10,11 ocular disease,^12,13 muscle dystrophy,¹⁴ and progeria,¹⁵ demonstrating the potential of BEs for the treatment of various genetic disorders.

Recently, prime editors (PEs) were developed for precise genome editing including all types of substitutions, small insertions, and deletions.¹⁶ PEs consist of Cas9 nickase fused with reverse transcriptase and a prime editing guide RNA (pegRNA) containing conventional gRNA, a primer binding sequence (PBS), and an RT template (RTT) sequence that encodes the desired edit. The PE-pegRNA complex binds the target site in the genome and nicks the opposite strand DNA, which is hybridized with PBS in the pegRNA. Then, a reverse transcriptase extends the 3′ end of DNA based on RTT in the pegRNA, and the extended DNA is incorporated into the genome. PEs expand the applicability and therapeutic potential of CRISPR nucleases and BEs.¹⁷

In this study, we first identified LNS-associated HPRT1 variants that are potentially targetable by BEs and PEs. We then applied BEs and PEs to (1) induce disease-associated mutations at the endogenous HPRT1 locus and (2) correct the targeted mutations as a therapeutic approach. Furthermore, we corrected HPRT1 pathogenic mutation in patient-derived fibroblasts and found that the functions of HPRT were restored by PE-mediated gene correction. Our findings collectively suggest a potential therapeutic strategy for LNS by using BEs and PEs to generate patient-specific disease models and correct the corresponding mutations.

Results

Analysis of HPRT1 variants associated with LNS

We first investigated the HPRT1 pathogenic variants known to be associated with LNS. A total of 615 genetic variants that cause either full or partial deficiency of HPRT activity were identified (http://www.lesch-nyhan.org), of which the majority were point mutations (381 variants, 62%) (Figure 1A). Among these point mutations, 226 (36.7%) were transition mutations and 155 (25.2%) were transversion mutations. Other HPRT1 variants were caused by deletion (158 variants, 25.7%) or duplication (40 variants, 6.5%). Of the 226 transition point mutations, 158 (69.9%) were C:G to T:A and 68 (30.1%) were A:T to G:C, which can be corrected by CBEs or ABEs, respectively. Theoretically, CRISPR-based PEs, which can induce all types of point mutations as well as small insertion or deletion (indel) mutations, can be used to correct 88% (541 variants) of the LNS-associated HPRT1 variants (Figure 1B).

Figure 1.

Analysis of LNS-associated HPRT1 variants that are targetable by BEs and PEs

(A) Classification of mutation types of HPRT1 variants derived from the LNS-associated genetic variants database (http://www.lesch-nyhan.org/). (B) The proportion of HPRT1 variants that are targetable by CRISPR-mediated CBEs, ABEs, and PEs.

Induction of patient-derived HPRT1 mutations using CBEs

We hypothesized that CBEs and ABEs could be used as paired systems. Because CBEs induce C:G to T:A conversion and ABEs induce A:T to G:C conversion, a C:G-to-T:A HPRT1 mutation can be introduced by CBEs in wild-type cells, and the HPRT1 mutation can be corrected by ABEs in mutated cells (Figure 2A). In other words, CBEs can induce pathogenic mutations in wild-type HPRT1 for disease modeling, and ABEs can correct pathogenic mutations in mutated cells. To demonstrate the effectiveness of the paired system, we selected two HPRT1 pathogenic mutations—NM_000194.3 c.430C>T and c.508C>T—that were identified in patients with LNS. We then carefully selected a target site for CBEs to minimize bystander alterations within active windows (i.e., approximately positions 4–8 from the 5′ end of the target sites) (Figure 2B; Table S2).

Figure 2.

Introduction and correction of patient-derived HPRT1 mutations using CBEs and ABEs

(A) Schematic overview of BE-meditated disease modeling and gene correction in human cells. (B) Target sequences of CBEs and ABEs for LNS-associated HPRT1 variants, c.430C>T (p.Q144∗) and c.508C>T (p.R170∗). The spacer sequences are indicated by boxes, and the target C:G pairs of CBEs and A:T pairs of ABEs are shown in blue and red, respectively. PAM sequences of each target site are shown in bold. (C) Heatmaps of CBE-mediated base editing frequencies for disease modeling of c.430C>T (top panel) and c.508C>T (bottom panel) in HEK293T/17 cells. Data are shown as means from two biologically independent samples. (D) Heatmaps of ABE-meditated base editing frequencies for correction of c.430C>T (top panel) and c.508C>T (bottom panel) in HEK293T/17 cells. Data are shown as the mean of three biologically independent samples. (E) Sanger sequencing results of endogenous c.430C>T (p.Q144∗) and c.508C>T (p.R170∗) target sites in mutated and corrected HAP1 cells. The red boxes indicate nucleotides converted by CBEs and ABEs. (F) Western blotting analysis of HPRT protein expression in mutated and corrected HAP1 cells. GAPDH was used as an internal control. (G) Crystal violet staining of mutated and corrected HAP1 cells selected in media containing 6-TG or HAT. (H) Results of IMP assay for HPRT activity in mutated and corrected HAP1 cells.

For the c.430C>T mutation, we selected the 430-#1 spacer sequence with an NGG PAM sequence that could be recognized by the conventional CBE with wild-type spCas9 as well as the engineered CBE with spCas9 variants with relaxed PAM constraints. Seven types of CBEs, including BE3, BE4max, AncBE4max, xCas9(3.7)-BE4, BE4max-NG, BE4max-SpG, and BE4max-SpRY, were tested in HEK293T/17 cells, and their base editing frequencies within the active windows were analyzed by targeted deep sequencing. Among the CBE variants, AncBE4max had the highest efficiency in converting the target C:T with a frequency of 10.3%, and the bystander C was also converted to T with a frequency of 1.4% (Figure 2C). We also found that the conventional CBE variants recognizing the NGG PAM sequences tend to have higher base editing activities compared with engineered CBE variants targeting alternative PAM sequences.

For the c.508C>T mutation, we chose the 508-#1 spacer sequence with an NG PAM sequence that could be recognized by the engineered CBE for PAM alteration (Figure 2B; Table S2). We tested four types of engineered CBEs including xCas9(3.7)-BE4, BE4max-NG, BE4max-SpG, and BE4max-SpRY and found that BE4max-SpG converted the target C:T with a frequency of up to 7.1% (Figure 2C). The bystander C was also converted to T at a frequency of 2.8%. Collectively, we successfully induced patient-derived HPRT1 mutations c.430C>T and c.508C>T into wild-type HPRT1 using CBE variants and found that AncBE4max and BE4max-SpG were the most efficient in introducing the two respective HPRT1 mutations.

Gene correction of patient-derived HPRT1 mutations using ABEs

Next, we attempted to correct the two patient-derived HPRT1 mutations using ABEs and additionally designed two gRNAs—430-#2 and 508-#2—on the complementary strands of each CBE target strand to correct the c.430C>T and c.508C>T pathogenic mutations (Figure 2B; Table S2). Both gRNAs had NG PAM sequences, and we tested four types of ABE variants—ABEmax-NG, ABEmax-SpG, ABEmax-SpRY, and ABEmax-xCas9(3.7). To measure the base editing efficiencies of these ABE variants, we established HEK293T/17 stable cell lines containing each target sequence of ABE (i.e., c.430C>T and c.508C>T) by lentiviral transduction. We transfected ABE variants with one of the two gRNAs into the HEK293T/17 stable cell lines and measured the base editing efficiency using targeted deep sequencing. As shown in Figure 2D, ABEmax-xCas9(3.7) corrected the c.430C>T HPRT1 mutation by up to 3% and 1.3% for the bystander A mutation, and ABEmax-SpG corrected the c.508C>T mutation by up to 5.2% without any bystander mutagenesis in the base editing active window. These results show that ABE could correct patient-derived HPRT1 mutations and that CBEs and ABEs could be used as paired systems in studying LNS-associated HPRT1 variants.

Disease modeling and gene correction of LNS-associated HPRT1 in HAP1 cell lines

For LNS disease modeling, we introduced endogenous HPRT1 mutations containing the c.430C>T or c.508C>T mutation into HAP1 cells, which are widely used for genetic model cell lines due to their haploidy characteristics.¹⁸ We first transfected two plasmids encoding CBE and either 430-#1 or 508-#1 gRNA to HAP1 cell lines and isolated single clones containing the intended endogenous HPRT1 mutations. Then, we delivered two plasmids encoding ABE and each 430-#2 and 508-#2 gRNA to these HPRT1 mutant clones and again isolated single clones bearing endogenous wild-type HPRT1 sequences. The mutated and corrected endogenous HPRT1 sequences of each clone were confirmed by Sanger sequencing (Figure 2E). To further analyze the off-target mutations in these HAP1 clones, we conducted in silico analysis of potential off-target sites of each gRNA using the Cas-OFFinder algorithm¹⁹ and targeted deep sequencing. We found a total of nine potential off-target sites in four HAP1 clones, and all clones had no detectable nucleotide substitutions and indels across any potential off-target sites (Figure S1; Table S3).

Next, we performed functional assessments to determine whether the genetically modified endogenous HPRT1 could significantly affect the functions of HPRT1. In a western blot assay for HPRT1, there was no detectable HPRT1 protein expression in c.430C>T- and c.508C>T-mutated HAP1 cells, whereas HPRT1 protein expression was recovered in c.430C>T- and c.508C>T-corrected HAP1 cells (Figure 2F). As 6-thioguanine (6-TG) is metabolized to a toxic compound by HPRT1, 6-TG can adversely affect the viability of cells expressing functional HPRT1, and hypoxanthine-aminopterin-thymine (HAT) medium selection can adversely affect the viability of cells with inhibited HPRT1 expression. We therefore treated c.430C>T-mutated HAP1 cells and c.508C>T-mutated HAP1 cells with 6-TG and HAT, respectively, and performed crystal violet staining to measure the cell viability. As shown in Figure 2G, c.430C>T- and c.508C>T-mutated HAP1 cells survived only in the 6-TG selection medium, and c.430C>T- and c.508C>T-corrected cells survived only in the HAT medium. We also performed a spectrophotometric assay using these HAP1 cells and found that c.430C>T- and c.508C>T-corrected cells could fully produce inosine monophosphate (IMP) from hypoxanthine (Figure 2H).

Assessment of LNS-associated HPRT1 mutations using PEs

We then examined if patient-derived HPRT1 indel mutations could be induced and corrected by PEs, which are able to induce mutations that cannot be induced by CBEs and ABEs. Based on our previous study on the clinical features and mutation spectrum of Korean patients with LNS,²⁰ we chose the insertional HPRT1 mutation, c.333_334ins(A), to test the feasibility of a PE-based therapeutic approach (Figure 3A). First, we optimized the PE system for HPRT1 c.333_334ins(A) induction and correction in HEK293T/17 cells. We chose three spacer sequences of PEs for inducing c.333_334ins(A) mutation into the wild-type HPRT1 sequence and designed 12 pegRNAs with variable lengths of PBS and RTT to introduce an adenine between the c.333 and c.334 sequences (Figures 3B; Table S2). The plasmids encoding PE2 and the pegRNAs were transfected to HEK293T/17 cells, and the prime editing frequencies were measured by targeted deep sequencing. We found that pegRNA-333-ins-#11 containing 15-nt PBS and 12-nt RTT induced the c.333_334ins(A) mutation by up to 3.4% (Figure 3C). Additional gRNAs for PE3 or PE3b systems were constructed to improve the prime editing efficiency, and we found that the prime editing efficiency was improved to 7.6% in the PE3b system (Figure 3D). As PE3 or PE3b systems could induce byproduct indel mutations at target sites, we also analyzed indel mutations excluding the wanted “A” insertional mutation and found that there were no significant byproduct indel mutations at target sites.

Figure 3.

Introduction and correction of patient-derived HPRT1 mutation using PE

(A) Schematic overview of PE-mediated disease modeling and gene correction in human cells. The patient-derived HPRT1 mutation, c.333_334ins(A), is representatively described. (B) Target sequences of PEs for the LNS-associated HPRT1 mutation, c.333_334ins(A) (p.G112Rfs∗10). The representative spacer-#1 sequence and PAM sequence are indicated in box and bold, respectively. The inserted adenine is highlighted in red. (C) Prime editing frequencies for introducing c.333_334ins(A) with pegRNAs containing variable lengths of PBS and RTT. The red arrow indicates the pegRNA used in subsequent experiments. (D) Prime editing frequencies of PE2, PE3, and PE3b to induce c.333_334ins(A) mutation. (E) Prime editing frequencies for correcting c.333_334ins(A) with pegRNAs containing variable lengths of PBS and RTT. The red arrow indicates the pegRNA used in subsequent experiments. (F) Prime editing frequencies with PE2 and PE3b to correct c.333_334ins(A) mutation. (G) Sanger sequencing results of endogenous c.333_334ins(A) (p.G112Rfs∗10) target sites in mutated and corrected HAP1 cells. The red arrow indicates the adenine inserted for disease modeling and gene correction by PEs. (H) Western blotting analysis of HPRT protein expression in mutated and corrected HAP1 cells. GAPDH was used as an internal control. (I) Crystal violet staining of PE-mediated mutated and corrected HAP1 cells selected with media containing 6-TG or HAT. (J) Results of IMP assay for HPRT activity in mutated and corrected HAP1 cells. Data are means from two or three biologically independent samples, and error bars indicate the standard error of the mean.

Next, we selected three additional spacer sequences of PEs for correcting the c.333_334ins(A) mutation (i.e., deleting the adenine between c.333 and c.334 sequences). A total of 15 pegRNAs (five pegRNAs per each spacer sequence) were designed to have variable lengths of PBS and RTT, and their prime editing efficiency was measured in HEK293T/17 cells containing lentivirus-transduced target sequences with c.333_334ins(A) mutation (Figure 3B; Table S2). The prime editing efficiency was dependent on the lengths of PBS and RTT, with maximal efficiencies reaching 10.5% with pegRNA-333-del-#5 (Figure 3E). We tried to further improve the prime editing efficiency by the PE3b system and found that pegRNA-333-del-#5 with PE3b system could improve the efficiency of adenine deletion to 46.7% (Figure 3F). Taken together, our results show that pegRNA-333-del-#5 could induce c.333_334ins(A) mutation by 5.1% and correct the mutation by 50% with the PE3b system in HEK293T/17 cells.

We also applied the PE3b system in HAP1 cell-based disease modeling and gene correction as shown in the CEB and ABE systems. We induced HPRT1 c.333_334ins(A) mutation with pegRNA-333-ins-#5 by the PE3b system, and single clones containing the intended mutation were isolated. We then corrected the mutations with pegRNA-333-del-#5 using the PE3b system and obtained single clones and examined the endogenous HPRT1 target sequences using Sanger sequencing (Figure 3G). Western blot assay showed that the expression of HPRT was disrupted in c.333_334ins(A)-mutant clones and restored in c.333_334ins(A)-corrected clones (Figure 3H). In line with the perturbation of protein expression, HPRT1 c.333_334ins(A)-mutant clones only survived in the 6-TG selective medium, and HPRT1 c.333_334ins(A)-corrected clones only survived in the HAT selective medium (Figure 3I). Spectrophotometric assay also showed that the production of IMP was distorted and recovered in HPRT1 c.333_334ins(A)-mutated and -corrected HAP1 cells, respectively (Figure 3J).

We then identified three potential off-target sites using in silico analysis, and the genomic DNA of these single clones was subjected to targeted deep sequencing. We found that there were no detectable off-target indel mutations in these sites (Figure S2; Table S3). Taken together, we demonstrated that PEs, as well as CBEs and ABEs, could be used for the functional assessment of LNS-associated HPRT1 variants.

Gene correction in LNS patient-derived primary cells

Lastly, we attempted to correct the HPRT1 c.333_334ins(A) mutation in LNS patient-derived fibroblasts. The fibroblasts were established from a 9-year-old Korean boy with LNS. The patient had global developmental delay, self-injurious behaviors (lip biting), hypotonia, and dystonia. His serum uric acid level was 9.3 mg/dL (reference range: 1.7–8.3 mg/dL). A renal ultrasound showed bilateral medullary nephrocalcinosis with a renal stone (Figure 4A). HPRT activity was null in erythrocytes. The patient was the second child of healthy non-consanguineous parents. His mother was a carrier, and his youngest brother was also diagnosed with LNS (Figure 4B).

Figure 4.

PE-mediated gene correction of c.333_334ins(A) mutations in patient-derived fibroblasts

(A) Renal ultrasound results of a patient with LNS with HPRT1 c.333_334ins(A) mutation. (B) Family pedigree of the patient with LNS with HPRT1 c.333_334ins(A) mutation. Fibroblasts were obtained from the patient with LNS as indicated with the red arrow. (C) Sequencing analysis of patient-derived fibroblasts to confirm the HPRT1 c.333_334ins(A) mutation. (D) Prime editing frequencies of various types of improved PEs and pegRNAs (top) for correcting the HPRT1 c.333_334ins(A) mutation. The red arrow indicates the pegRNA used in the subsequent experiment. Representative results of high-throughput sequencing of patient-derived fibroblasts treated with PE5max and tevopreQ pegRNA to correct the HPRT1 c.333_334ins(A) mutation (bottom) are shown. (E) Crystal violet staining of PE-mediated corrected fibroblasts with media containing 6-TG or HAT. Data are means from two biologically independent samples, and error bars indicate the standard error of the mean. (F) Western blotting analysis of HPRT protein expression in patient-derived fibroblast cells and HPRT1 c.333_334ins(A)-corrected patient-derived fibroblasts selected with HAT medium. GAPDH was used as an internal control.

We first confirmed the endogenous HPRT1 sequences by Sanger sequencing and found that the patient-derived cells indeed had the c.333_334ins(A) mutation (Figure 4C). To correct the HPRT1 mutation, we electroporated plasmids encoding the pegRNA-333-del-#5 with the PE3b system, which showed the highest gene correction efficiency in HEK293T/17 cells (46.7%), into patient-derived primary cells; however, we found that HPRT1 gene correction occurred in only 0.4% of these cells (Figure 4D).

Recently, two studies showed that enhanced prime editing systems have improved prime editing efficiency. Nelson et al.²¹ reported that the incorporation of structured RNA motifs to the 3′ end of pegRNAs can improve their ability to install intended mutations, and Chen et al.²² reported that the inhibition of mismatch repair pathway using double-negative MLH1 protein with an optimized PE, PEmax, enhanced the prime editing efficiency. To confirm whether the engineered pegRNAs could be applied to LNS patient-derived primary cells, we cloned two structured RNA motifs—tevopreQ1 and tmpknot—to the 3′ end of pegRNA-333-del-#5 and tested them using three improved PEs—PE2max, PE3max, and PE5max. As shown in Figure 4D, we found that the pegRNA with tevopreQ1 scaffold induced adenine deletion more efficiently than the pegRNA with tmpknot scaffold, and pegRNA-333-del-#5 with tevopreQ1 scaffold showed an adenine deletion efficiency of 14% in the PE5max system. The corrected endogenous HPRT1 sequences were also confirmed by targeted deep sequencing. Crystal violet staining of 6-TG and HAT medium-selected cells showed that HPRT1-corrected primary cells were viable in HAT medium selection, and HPRT protein expression was also confirmed in these HAT medium-selected fibroblasts (Figures 4E and 4F).

Discussion

We demonstrated that CBEs and ABEs can be used as a paired system to induce and correct c.430C>T and c.508C>T mutations and that PEs can be used to study the c.333_334ins(A) HPRT1 mutation in depth. Importantly, we showed that PEs can correct the HPRT1 mutation in patient-derived fibroblasts by up to 14% and that the expression of HPRT is confirmed in HAT medium-selected fibroblast cells. Our current study is the first to apply the CRISPR system for gene therapy in LNS. We successfully established cellular models of LNS containing patient-derived HPRT1 mutations and corrected these mutations using CRISPR-mediated BEs and PEs. We used CBEs to induce pathogenic mutations, c.430C>T and c.508C>T, in wild-type HAP1 cells and established cellular models bearing these HPRT1 mutations. Then, we successfully corrected these mutations using ABEs, which demonstrated the potential of BEs for use in gene therapy. We further applied PEs to induce and correct the patient-derived HPRT1 mutation, c.333_334ins(A), in HAP1 cells and patient-derived fibroblasts and found that the functions of HPRT can indeed be restored in HPRT1 gene-corrected cells.

BEs are powerful tools for introducing nucleotide conversions in the genome. BEs can induce nucleotide transition mutations such as C:G-to-T:A or A:T-to-G:C conversions, which account for 36.7% of LNS-associated HPRT1 mutations; yet, bystander nucleotide substitutions may occur because BEs can generate any base transition within the active window. To reduce the frequency of unwanted mutations, we carefully chose the target sites and used engineered BE variants with PAM flexibility, such as xCas9, SpG, Cas9-NG, and SpRY variants. By doing so, we were able to induce and correct LNS-associated mutations without unwanted mutations. To compare the performance of BE variants, we induced nine additional HPRT1 mutations in HEK293T cells. We found that AncBE4max showed the highest C:G-to-T:A conversion across five target sites containing NGG PAM, but other BE variants showed no trend of base editing activity across target sites (Figure S3). Thus, optimizing BE systems for target mutations is important becuase base editing activity is affected by many factors such as PAM preference of BE variants, nucleotide contexts around the target cytosine or adenine, positions of target cytosine or adenine at base editing window.

PEs enable precise genome editing, including all types of nucleotide conversions as well as small insertion and deletions; however, the prime editing activities of PEs are cell-type dependent compared with those of Cas9 and BEs. Recently, Chen et al.²² reported that the inactivation of MLH1, which is related to the DNA mismatch repair pathway, enhances the prime editing activity, and Nelson et al.²¹ showed that engineered pegRNA (epegRNA) improves prime editing efficiency by stabilizing the pegRNA in HEK293T/17 cells. As the PE5 system and epegRNA show higher prime editing efficiency than the original PE system and pegRNA for many cell types, they are worthwhile to be used first when trying prime editing in primary cells. In this study, we also found that PEs had up to 50% gene correction efficiency for the HPRT1 c.333_334ins(A) mutation in HEK293T/17 cells. However, the original PE system could not efficiently correct the mutation in patient-derived fibroblasts, and even the engineered PE systems (epegRNAs and PEmax) could only correct HPRT1 c.333_334ins(A) mutations by up to 14%; the efficiency of PEs should be further improved in order to be used in therapeutic approaches of LNS.

HPRT enzyme deficiency causes uric acid overproduction, causing hyperuricemia and gout. However, the physiopathology underlying behavioral abnormality, intellectual disability, and self-harm in LNS has been unknown, and there are no fundamental treatment modalities for LNS.²³ While gout can be treated by controlling the uric acid levels with allopurinol, neurological or behavioral symptoms are totally unaffected by allopurinol.²⁴ Although several animal models have been developed to test therapeutic agents for metabolic abnormalities consistent with deficient purine salvage systems,^25,26 HPRT1-knockout animal models failed to recapitulate neurological alterations in LNS such as dystonia and self-injurious behaviors.^27,28 The mechanisms of LNS phenotypes can be studied using cell models such as patient-derived peripheral blood lymphocytes, patient-derived fibroblasts, mouse Hprt1-knockout neuronal cells, and human HPRT1-knockout embryonic carcinoma, but they are not sufficient to study the systematic phenotypes of LNS because various types of cell lines with HPRT deficiency cannot perfectly reproduce the complicated phenotypes of LNS.^29,30,31 Therefore, our results of established cell lines using CRISPR-Cas9 could overcome the limitations of study using existing cell models and animal models to develop more fundamental treatments, but more advanced research is needed in order to improve the efficiency of gene correction.

CRISPR-mediated BEs and PEs are widely used for precise genome engineering in living systems. By allowing the introduction and correction of disease-associated mutations at the target locus of the identical genetic background, these systems are useful for adequately performing functional studies on patient-derived mutations and developing personalized treatments. Our current results show that CRISPR-mediated BEs and PEs have the potential to be used as powerful tools in studying LNS-associated HPRT1 variants.

Materials and methods

Classification of HPRT1-related variants in LNS-associated database

A total of 615 LNS-associated variants were identified through the HPRT1 mutation database on the official website of the Lesch-Nyhan Disease International Study Group (http://www.lesch-nyhan.org/). The genetic position of each mutation was obtained from the National Center for Biotechnology Information (NCBI) website and classified according to the mutation types. The coverage rate of CBE and ABE was calculated from the number of T:A-to-C:G point mutations and the number of G:C-to-A:T point mutations from the wild-type HPRT1 sequences, respectively. PE-editable mutations including all point mutations, insertions (less than 40 bp), and deletions (less than 80 bp) were counted as previously described.³²

Plasmid construction

pRG2 was used for gRNA cloning, and pU6-pegRNAGG-acceptor, pU6-tevopreq1-GG-acceptor, and pU6-tmpknot-GG-acceptor plasmids were used for pegRNA cloning. To construct lentiviral vector-containing HPRT1 mutant target sequences for generating HEK293T/17 stable cell lines, HPRT1 mutant target sequences were cloned into lentiGuide-Puro plasmids. Detailed information on plasmids is provided in Table S1, and the sequences of target sites, PBS, and RTT are listed in Table S2.

Isolation of patient-derived fibroblasts

The experiments using patient fibroblasts were approved by the Institutional Review Board of Asan Medical Center (approval #2017-1347). A 3 mm round skin biopsy was manually divided into 5–6 pieces with sharp edges using a scalpel in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin. The samples were attached to the bottom of the flask by gently tapping or a sliding motion. After 15 min, 10 mL medium was added to the T25 flask for culture. At confluency, the fibroblasts were rinsed with Dulbecco’s PBS and trypsinized using Trypsin/EDTA solution (Gibco). The medium was changed every 3 to 4 days.

Cell culture and cell line generation

HEK293T/17 cells (ATCC CRL-11268) were maintained in DMEM, and HAP1 cells were maintained in Iscove’s modified Dulbecco’s medium supplemented with 10% FBS and 1% penicillin-streptomycin. To generate stable cells containing HPRT1 mutant target sequences, lentivirus was produced and transduced into HEK293T/17 cells. Briefly, 2 × 10⁵ HEK293T/17 cells were seeded onto 24-well plates and transfected with 1 μg plasmid DNA (500 ng viral vector-containing HPRT1 mutant sequences, 300 ng psPAX2, and 200 ng pMD2.G) using Lipofectamine 2000 (Thermo Fisher Scientific) according to the manufacturer’s instruction. The medium was changed 24 h after transfection, and the supernatant was harvested 48 h after transfection and filtered using a 0.45 μm filter. The lentivirus soup was stored at −70°C until use. HEK293T/17 cells were transduced with the lentivirus and selected for 3 days using 1 μg/mL puromycin.

Transfection and drug selection

For plasmid transfection, 1.5 × 10⁵ HEK293T/17 cells or 0.8 × 10⁵ HAP1 cells were seeded onto a TC-treated 24-well plate 1 day before transfection, and transfection was conducted when cell confluency reached 60%–70%. A total of 2 μg plasmids (1.5 μg BEs with 500 ng gRNAs or 1.5 μg PEs with 500 ng pegRNAs) were delivered using 3 μL Lipofectamine 2000 (Thermo Fisher Scientific) according to the manufacturer’s protocol. For the PE3 and PE3b experiments, 166 ng plasmids encoding gRNAs were additionally delivered to the cells. To correct the HPRT1 c.333_334ins(A) mutations in LNS patient-derived fibroblasts, 1 × 10⁵ patient-derived fibroblasts were pulsed twice with 20 ms width and 1,200 voltage using Neon Transfection System 10 μL Kit (Thermo Fisher Scientific). For the functional assessment of HPRT1, cells were selected for 10 days in 10 μg/mL 6-TG (Sigma-Aldrich) and a medium supplemented with HAT Media Supplement (50×) Hybri-Max (Sigma-Aldrich).

Sanger sequencing and targeted deep sequencing

To confirm the endogenous HPRT1 sequences of single clones, genomic DNA was extracted using the DNA Blood & Tissue kit (Qiagen) according to the manufacturer’s protocol, and target sites were amplified using Phusion High-Fidelity DNA Polymerase (New England Biolabs). The PCR products were purified using QIAquick PCR Purification Kit (Qiagen) and subjected to Sanger sequencing. To analyze the mutation frequencies by targeted deep sequencing, the target sites were amplified using Phusion High-Fidelity DNA Polymerase (New England Biolabs) with Illumina TruSeq HT dual index adaptor primer, and the libraries were sequenced at 150 bp paired end using the Illumina MiniSeq or iSeq 100 sequencing equipment. Two web-based analytic tools—Cas-Analyzer (http://www.rgenome.net/cas-analyzer)³³ and BE-Analyzer (http://www.rgenome.net/be-analyzer/)³⁴—were used to analyze the sequencing data. The primer sequences are listed in Table S3.

Crystal violet staining

For crystal violet staining, cells were seeded onto 24- or 48-well plates and maintained in a medium containing 6-TG or HAT. Ten days after drug selection, media were removed, and the cells were fixed with 4% paraformaldehyde. The fixed cells were stained with 1% crystal violet staining solution (Sigma Aldrich) for 20 min and washed with distilled water, and the images of each well were captured.

Western blot assay

For the isolation of total protein, harvested cells were resuspended in the PRO-PREP protein extraction solution (Intron Biotechnology). Protein concentrations were quantified by Bradford protein assay (Bio-Rad). Total proteins were separated on 12% SDS-PAGE gel and transferred to a nitrocellulose membrane (Bio-Rad). The membranes were blocked with 4% skim milk in Tris-buffered saline (TBS) containing 0.1% Tween 20 (TBST; 10 mM Tris-HCL [pH 7.5], 150 nM NaCl, and 0.1% Tween 20). The blocked membranes were incubated with the primary antibodies against HPRT (Abcam, 1:1,000) or GAPDH (Enzo Life Sciences, 1:5,000) at 4°C overnight. After washing with TBST, the samples were incubated with HRP-conjugated secondary antibody (Cell Signaling, 1:10,000) in TBST containing 4% skim milk at room temperature for 1 h. After washing the membrane, signals were detected using an enhanced chemiluminescence (ECL) system (Thermo Fisher Scientific) according to the manufacturer’s protocol.

IMP assay

The activity of HPRT was measured in sonicated lysates of cultured cells using the PRECICE HPRT Assay kit (Novocib). The cells were washed two times with PBS, trypsinized, and centrifuged. The resulting pellets were resuspended in 1 mL water, sonicated for 1 min, and centrifuged at 13,000 RPM for 15 min to clarify the lysate. After protein concentration, 100 μL (total protein amount: 100 μg) cell extracts were placed in a well, followed by the addition of 100 μL 2× concentrated reaction buffer (from PRECICE HPRT kit) with or without 2 mM PRPP. Reaction was followed by measuring absorbance at 340 nm in 5 min intervals for 2 h.

Data availability

DNA sequencing data have been deposited in the NCBI Sequence Read Archive (SRA) database with BioProject accession code SRA: PRJNA868858.