CRISPR-Mediated Genomic Addition to CPS1 Deficient iPSCs is Insufficient to Restore Nitrogen Homeostasis.

CPS1 deficiency is an inborn error of metabolism caused by loss-of-function mutations in the CPS1 gene, catalyzing the initial reaction of the urea cycle. Deficiency typically leads to toxic levels of plasma ammonia, cerebral edema, coma, and death, with the only curative treatment being liver transplantation; due to limited donor availability and the invasiveness and complications of the procedure, however, alternative therapies are needed. Induced pluripotent stem cells offer an alternative cell source to partial or whole liver grafts that theoretically would not require immune suppression regimens and additionally are amenable to genetic modifications. Here, we genetically modified CPS1 deficient patient-derived stem cells to constitutively express human codon optimized CPS1 from the AAVS1 safe harbor site. While edited stem cells efficiently differentiated to hepatocyte-like cells, they failed to metabolize ammonia more efficiently than their unedited counterparts. This unexpected result appears to have arisen in part due to transgene promoter methylation, and thus transcriptional silencing, in undifferentiated cells, impacting their capacity to restore the complete urea cycle function upon differentiation. As pluripotent stem cell strategies are being expanded widely for potential cell therapies, these results highlight the need for strict quality control and functional analysis to ensure the integrity of cell products.

Introduction

Carbamoyl phosphate synthetase 1 (CPS1, EC 6.3.4.16) deficiency (OMIM #237300) is a rare inborn error of metabolism affecting the first, rate-limiting enzyme of the urea cycle, the hepatocyte function in terrestrial mammals that incorporates waste ammonia into urea for excretion; this disorder affects 1 in 300,000 to 1 in 1.3 million live births worldwide [1,2]. Functional impairment or loss of CPS1 typically results in elevated plasma ammonia without orotic aciduria, leading to nausea, lethargy, encephalopathy, and coma [3]. These symptoms rapidly progress and may lead to death despite rapid diagnosis and aggressive treatments including hemodialysis and ammonia scavenger drugs [4-6]. However, even if these treatments are successful in mitigating acute plasma ammonia elevations, they are incapable of preventing recurrent crises and the irreversible neurocognitive damage associated with them.

As a monogenic disorder with well-defined biochemistry, CPS1 deficiency represents a seemingly prime target for classic gene therapy approaches. Recent gene therapy successes for all the other urea cycle enzymes [7-10] also encourage this approach, but CPS1 has multiple unique hurdles that make implementing such strategies difficult, including relatively high therapeutic threshold [11] and relatively large cDNA size. These challenges were recently overcome using helper-dependent adenovirus- [12] and dual adeno-associated virus (AAV)-based [13] gene replacement approaches, both demonstrating that lost murine Cps1 expression could be replaced with ectopic expression of murine or human CPS1, respectively. Though successful, the adenoviral approach is limited by potential immune complications [14]; in addition, the dual AAV strategy is at present somewhat limited by high viral loads that may be impractical for production and patient administration. Both viral strategies may be further impacted by waning efficacy over time due to episomal loss.

Cell-based therapies for metabolic disorders in general, and CPS1 deficiency in particular, may provide a viable alternative to virus-based gene therapies. Liver transplantation is generally curative but requires long-term immune suppression and faces continued scarcity in supply of donor organs [3]; in addition, patients often still require dietary supplementation of citrulline (a key urea cycle intermediate) [15]. Patient-derived induced pluripotent stem cells (iPSCs) could circumvent the issues associated with finding compatible donor organs as well as viral delivery of genetic materials. Successful cell replacement with exogenous hepatocytes has been reported in some studies for metabolic disorders [16-18], including CPS1 deficiency [19], though rigorous and well-controlled trials remain to be performed. Human hepatocyte transplantation in an immunosuppressed murine model has previously been shown to treat another urea cycle disorder, arginase 1 deficiency [20], effectively, and iPSC-derived hepatocytes are beginning to show similar promise in some enzyme activities when compared to primary hepatocytes [21]. An important advantage of iPSC-derived hepatocyte-like cells (HLCs) is that cell quality can be controlled prior to transplantation, potentially generating a functionally unlimited supply of hepatocytes to replace the endogenous dysfunctional cells.

The discovery and implementation of the clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9 gene editing platform [22,23] has revolutionized genome editing in recent years, providing a rapid and reliable tool to make precise modifications. Multiple types of mammalian cells have been edited with CRISPR/Cas9, including human iPSCs [24]. One advantage of using patient-derived iPSCs for in vitro editing, as opposed to primary cells ex vivo, is that targeted changes can be rigorously validated by sequencing prior to transplantation; in addition, potential off-target cleavage events can be predicted and mitigated through careful design, and also subsequently confirmed. iPSCs edited with CRISPR/Cas9 have been shown to robustly express transgenes [25], including those of urea cycle enzymes [26], making this an attractive option for CPS1 deficiency.

Herein, we reprogrammed patient-derived CPS1 deficient fibroblasts to iPSCs and subsequently utilized CRISPR/Cas9 to introduce ectopic CPS1 expression from the AAVS1 safe harbor locus in a genomic addition approach. The resultant cells were then differentiated to HLCs to assess their ammonia metabolizing capability.

Results

Reprogramming of CPS1 Deficient Fibroblasts Successfully Produces Pluripotent Stem Cells

To establish a human cell model for CPS1 deficiency, we acquired three deidentified patient fibroblast lines, termed CD1, CD2, and GM03432 (the latter subsequently referred to as GM) (Figure 1A). Clinical reports from all three patients indicated each had neonatal onset of disease symptoms; CD1 cells are from an afflicted male, while CD2 and GM cells are from afflicted females. CD1 cells contain a homozygous nonsense mutation resulting from the transversion c.2494G>T, leading to amino acid residue 832 converting from glutamate to a stop codon (E832X). CD2 is compound heterozygous with c.2740G>C, changing aspartate 914 to histidine (D914H), and a splice mutation resulting in exon 36 skipping (c.4162-2A>G). GM also contains a homozygous mutation, but it is missense c.610G>T, resulting in a valine to phenylalanine transition at residue 204 (V204F).

Figure 1

Characterization of patient-derived iPSCs. A. Table detailing the mutations in the three cell lines. CD1 and GM are both homozygous, while CD2 is compound heterozygous. B. Bright field images of patient-derived parental fibroblasts and their corresponding daughter iPSCs. C. Fold change qPCR analysis of pluripotency genes in patient iPSCs compared to the human embryonic stem cell line H9 (normalized to 1). Data are mean ± SEM, n = 3 per group. P values are from one-way ANOVA. D. Immunocytochemistry of pluripotency markers in iPSCs compared to H9 cells. E. H&E staining of sections from teratomas formed by each cell line. All three lines generate tissues derived from the three primary germ layers. F. Karyotype analysis of CD1 (top left panel), CD2 (top right panel), and GM (bottom left panel) iPSCs. Red box in CD1 indicates balanced chromosomal translocation between chromosomes 19 and 20.

These three lines were reprogrammed using a lentivirus expressing the four classic Yamanaka factors OCT4, SOX2, KLF4, and MYC, and clones from each cell line were used for characterization. All iPSCs exhibited typical morphology (Figure 1B), and gene expression analysis demonstrated that they express the pluripotency genes OCT4, SOX2, NANOG, REX1, and TERT at similar levels to the established human embryonic stem cell line H9 (Figure 1C), chosen as a well-characterized reference of pluripotency. Immunocytochemistry showed that these cells express the pluripotency markers OCT4, KLF4, TRA-160, and TRA-181 at the protein level (Figure 1D). When injected into immunocompromised mice, cells were also capable of forming teratomas with tissue derived from all three germ layers (Figure 1E). Karyotype analysis demonstrated that CD2 and GM iPSCs have no detectable large chromosomal aberrations (Figure 1F); however, CD1 was found to have a balanced translocation between the short arm of chromosome 19 and the long arm of chromosome 20 (red box). This mutation was also identified in the parental fibroblasts, however; thus, this was not a result of reprogramming. A single clone from each line was chosen for all downstream experiments.

CRISPR/Cas9-Mediated Genomic Addition Successfully Inserts Human Codon Optimized CPS1 into the AAVS1 Safe Harbor Locus

To restore expression in CPS1 deficient cell lines, human codon-optimized CPS1 cDNA (hcoCPS1) was introduced into the AAVS1 safe harbor locus using two guide RNAs and Cas9 nickase as in the previously described AFP reporter cell (ARC) vector [27] with modifications. Here, the human EF1α promoter drives the expression of hcoCPS1, which is terminated by the human growth hormone polyadenylation sequence. An in-frame puromycin resistance gene allows for drug selection of transfected cells after integration, with the entire cassette flanked by homology arms to AAVS1. A schematic of the genome editing strategy is depicted in Figure 2A.

Figure 2

Design and validation of CRISPR/Cas9 editing in iPSCs at the . A. Schematic of the overall CRISPR/Cas9 strategy. B. Gel images of PCRs amplifying either the 5’ or 3’ junctions, defined by the bracketed lines in (A). Primer locations are indicated by black arrows (5’ junction) and blue arrows (3’ junction) in (A). C. Sequencing of the 5’ and 3’ junctions showing faithful insertion of the transgenic construct.

After integration, puromycin-resistant cells were subcloned, and genomic DNA was extracted to determine if integration had been faithfully completed. PCR analysis of both the 5’ and 3’ junctions was performed using primer pairs amplifying from outside of the homology arms and inside the transgenic cassette (Figure 2A, black and blue arrows); the expected bands were present in edited cells but not in unedited cells (Figure 2B). Junction PCR products were subsequently sequenced, demonstrating seamless integration of the expected genomic region and transgene sequences without errors (Figure 2C).

Differentiation of iPSCs Resulted in Hepatocyte-Like Cells Inefficient in Ammonia Metabolism

To determine the capacity of these iPSCs to form hepatocyte-like cells (HLCs), iPSCs were differentiated using a previously described method [28]. Figure 3A depicts the overall differentiation strategy. After a 4-day endoderm induction, cells were incubated for 8 days with hepatocyte growth factor (HGF) and dimethylsulfoxide (DMSO) to induce hepatic specification, followed by 6 days in dexamethasone, an extension of 3 days from the original protocol that modestly improves urea cycle gene expression [27]. The subsequently produced HLCs were maintained for 3 days in the presence of dexamethasone, hydrocortisone, insulin, and DMSO before being harvested for analysis on day 21. Day 21 HLCs from all three lines, both unedited and edited, exhibited the characteristic hepatocyte cobblestone morphology (Figure 3B). Edited cells did not statistically significantly differ from unedited cells in maturity based on expression of the immature hepatocyte gene α-fetoprotein (AFP; p = 0.5) and mature genes cytochrome p450 3A4 (CYP3A4; p = 0.3) and fumarylacetoacetate hydrolase (FAH; p = 0.8) (Figure 3C); additionally, edited and unedited cells both secreted albumin at comparable levels (1095ng/day/million cells ± 507 vs. 595ng/day/million cells ± 377 [unedited vs. edited, respectively]; p = 0.4) (Figure 3D). To measure the ammonia consumption capacity of differentiated HLCs, NH4Cl was added to cell media at a final concentration of 2.5mM; supernatants were harvested after 24 hours and assayed to measure ammonia levels. Unexpectedly, unedited cell supernatants contained significantly less ammonia than their edited counterparts as a whole (unedited: 539.1µM ± 32.0; edited: 1215.0µM ± 64.3; p < 0.0001) and within each individual cell line (CD1: 539.3µM ± 9.05 vs. 1302µM ± 56.8, p < 0.001; CD2: 612.6µM ± 81.1 vs. 990.5µM ± 89.8, p < 0.05; GM: 465.3µM ± 14.6 vs. 1351µM ± 8.4, p < 0.0001) (n = 3 for each cell line, total of n = 9 per group [unedited or edited]) (Figure 3E). Urea levels were below the limits of detection (data not shown).

Figure 3

Differentiation of iPSCs to HLCs. A. Schematic diagram of the hepatic differentiation strategy. iPSCs are directed progressively through the definitive endoderm, hepatic specification, and maturation stages, using the listed growth factors to reach the HLC state. B. Representative bright field images of day 21 HLCs from all three lines with and without CRISPR/Cas9 genome editing. No salient morphological differences were observed. C. Fold change qPCR analysis of unedited (normalized to 1) and edited day 21 HLCs. Expression of the less mature AFP and more mature CYP3A4 and FAH genes are not significantly different between the two. D. Levels of albumin secreted into cell culture media of day 21 HLCs after 24 hours. E. Levels of ammonia found in the supernatant of day 21 HLCs after 24 hours of treatment with 2.5mM NH4Cl. Data in C-Eare mean ± SEM, n = 9 per group; n = 5 in edited group of (D).

Investigating Potential Mechanisms of Differential Ammonia Metabolism

To address the unexpected results from the in vitro ammonia challenge, RNA was extracted from HLCs and subjected to qPCR analysis for the 6 enzymes and 2 transporters of the urea cycle, in addition to hcoCPS1, to further assess function (Figure 4A). There were no significant differences in gene expression between unedited and edited HLCs (n = 3 per cell line, total n = 9 per group) except for endogenous CPS1 (0.52-fold ± 0.15 decrease in edited cells; p = 0.017). Unexpectedly, hcoCPS1 expression was not markedly elevated in edited cells (1.82-fold ± 0.4; p = 0.22), prompting an investigation into the cause of this poor expression. Loss of hcoCPS1 expression was not due to heterochromatin formation as these cells retained puromycin resistance (data not shown), suggesting some other mechanism. DNA methylation is one method of gene regulation; cytosine residues in CpG dinucleotides are covalently methylated at the fifth position of the pyrimidine ring to form 5-methylcytosine. This event is commonly associated with transcriptional silencing [29]; global DNA methylation status is known to fluctuate during differentiation [30]. To determine the DNA methylation status at the transgene promoter in HLCs, genomic DNA was extracted and subjected to bisulfite conversion. Bisulfite conversion deaminates cytosine into uracil but has no effect on 5-methylcytosine. Subsequent sequencing can demonstrate if cytosines were protected from conversion by a methyl group, providing insight into local DNA methylation states. All three edited lines demonstrated methylation at all 19 CpG dinucleotides in the transgenic EF1α promoter, in addition to 7 CpGs in the flanking DNA (2 of them upstream and the remaining 5 downstream of the promoter) (Figure 4B). Unexpectedly, promoter methylation was not a result of differentiation as undifferentiated iPSCs also showed complete methylation at all CpGs (Figure 4C). The donor plasmid used during nucleofection showed no methylation (Figure 4C), indicating that promoter methylation was iPSC-derived after integration. Loss of hcoCPS1 expression was confirmed in iPSCs; shortly after CRISPR/Cas9 editing and puromycin selection, iPSCs showed significantly increased expression of hcoCPS1 (26.02-fold ± 11.16 compared to unedited; p = 0.040) (Figure 4D) which was lost upon subsequent culturing (1.76-fold ± 0.68 compared to unedited; p = 0.7) (Figure 4E). Transcriptional silencing was accompanied by undetectable levels of CPS1 enzyme activity in edited iPSCs (data not shown).

Figure 4

Investigating the factors contributing to differential ammonia metabolism in unedited and edited HLCs. A. Fold change qPCR of urea cycle-related genes (6 enzymes and 2 transporters) in day 21 HLCs, where unedited cell samples are normalized to 1. B and C. DNA methylation analysis of edited day 21 HLCs (B) and edited iPSCs with donor plasmid control (C). Unfilled circles represent unmethylated CpGs, while filled circles represent methylated CpGs. D and E. Fold change qPCR analysis of hcoCPS1 in unedited (normalized to 1) compared to edited iPSCs after only a few passages (D) and several passages (E). F. Fold change qPCR analysis of glutamate/glutamine metabolic genes in day 21 HLCs, where unedited cell samples are normalized to 1. Data in (A), (D), (E), and (F) are mean ± SEM, n = 9 per group.

While promoter methylation explains the loss of hcoCPS1 expression, it fails to account for the difference in ammonia reduction (Figure 3D) between unedited and edited HLCs. Hepatocytes in vivo are central in regulating amino acid metabolism [31], including those whose synthesis involves the direct incorporation of ammonia. The reversible production of glutamine involves the condensation of glutamate with ammonia, regulated by glutamine synthetase (GLUL) and glutaminase 2 (GLS2). In addition, glutamate itself is reversibly produced from ammonia and α-ketoglutarate via the action of glutamate dehydrogenase (GLUD1). Because hcoCPS1 expression is lost in edited cells, we hypothesized that differential expression of any of these three metabolic genes could account for varying levels of ammonia utilization. However, qPCR analysis showed that none of these three genes were significantly differentially expressed between unedited and edited HLCs (GLUL p = 0.1; GLS2 p = 0.2; GLUD1 p = 0.6) (Figure 4F).

Discussion

CPS1 deficiency is an inborn error of metabolism in which ureagenesis stalls due to lack of input carbamoyl phosphate from CPS1 enzymatic activity; waste nitrogen does not flux through the cycle thus preventing its eventual conversion to non-toxic urea that is excreted in urine. Symptoms typically arise in the neonatal period, though onset may occur at any time in childhood or into adulthood depending on the extent of protein dysfunction, with as many as 50% of neonatal cases leading to mortality [1]. Dietary protein restriction and ammonia scavenger drug administration are the mainstays of current therapy, but they are incompletely effective and have remained largely unchanged since the 1980s; orthotopic liver transplantation is generally curative, though donor organs remain scarce and the procedure is accompanied by lifelong immune suppression and its associated complications and risks; in some cases patients still require additional citrulline supplementation as small intestinal enterocytes remain CPS1 deficient [15]. Currently, there are no treatments that fully address the underlying biochemical imbalance or reverse the accumulated neurocognitive deficits from repeated episodes of hyperammonemia.

To develop a novel therapeutic option for CPS1 deficiency patients, we reprogrammed patient-derived fibroblasts into iPSCs and inserted human codon optimized CPS1 cDNA into the AAVS1 locus for constitutive expression via the EF1α promoter using CRISPR/Cas9. Edited iPSCs were differentiated to HLCs to determine their ammonia metabolizing capacity relative to their isogenic, unedited controls. Upon challenging with NH4Cl, edited HLCs removed significantly less ammonia from the supernatant than unedited HLCs. This was an unexpected outcome as overexpression of hcoCPS1, in the presence of similar expression levels of the other urea cycle enzymes, would be expected to increase ammonia utilization. However, upon further investigation, hcoCPS1 expression was found to be silenced due to promoter methylation.

The concept of using a safe harbor for genomic addition is not new; AAVS1-based transgene insertion and expression is a long-standing concept in stem cell biology and has many reported successes [32,33]. However, transgene silencing mediated by promoter methylation has also been reported in undifferentiated [34] and differentiated [32] cells. The present and prior studies suggest that promoter choice is a critical component in avoiding silencing. Previous work from our group [26] and others [35] has demonstrated that the EF1α promoter expresses well in PSCs, suggesting some other factor remains to be elucidated and is impacting methylation status. Cell-line specific alterations, and potentially global DNA methylation changes, may also exist that could explain the difference in unedited and edited cells. Therefore, preventing DNA methylation-dependent silencing, regardless of promoter choice, will need to be optimized prior to advancing iPSC-derived HLCs as a putative treatment option. This could be achieved using universal chromatin opening elements [36] or small molecule inhibitors of DNA methyltransferases [37], some of which are already FDA approved. Determining the causal factors of promoter silencing, and establishing innovations that resolve them, will be important in moving transgenic PSC technologies towards more translatable therapies.

The difference in ammonia utilization in unedited and edited HLCs, despite no obvious urea cycle-related gene expression differences (except for endogenous CPS1, the primers for which may have inefficient but non-negligible binding to hcoCPS1), suggests that unedited HLCs shuttled ammonia into an alternative pathway. In support of this notion, no difference was observed in urea formation between the two populations (data not shown) in preliminary experiments when treated with heavy isotope-labeled 15NH4. In addition to urea, ammonia may be condensed with α-ketoglutarate to form glutamate by the enzyme glutamate dehydrogenase (GLUD1, EC 1.4.1.3). Glutamate may then receive an additional ammonia moiety to form glutamine via glutamine synthetase (GLUL, EC 6.3.1.2). Both reactions are reversible either by GLUD1 itself or by the action of glutaminase 2 (GLS2, EC 3.5.1.2), respectively. Together, these three enzymes represent a small portion of urea cycle-independent enzymes that directly or indirectly affect ammonia levels in hepatocytes, making them attractive targets to investigate as potential differential regulators of ammonia levels in unedited and edited HLCs. However, gene expression levels were not significantly different between the two cell populations, suggesting that if they are playing a role, then this is likely at the post-transcriptional level. Metabolic flux through these enzymes can be investigated using metabolomics approaches with 15NH4 or other isotopic nitrogen-containing compounds [38]; this may also clarify the roles of other processes affecting ammonia indirectly, such as amino acid uptake, in an unbiased way.

While there is great potential for iPSC-derived HLCs as a therapeutic, they do have important disadvantages. One significant drawback of utilizing iPSC-based technology for cell replacement is the time needed to generate, edit, and differentiate iPSCs to the desired fate. In metabolic disorders such as CPS1 deficiency, diagnosis can be made prenatally by amniocentesis [39] or chorionic villi biopsy [40], taking advantage of a key time window during which the unborn fetus relies on maternal ammonia clearance. This approach would potentiate the development of a cell-based therapeutic tailored to the patient that is ready for administration at or soon after birth. Recent advances in liver organoids [41] and scaffolding for transplantation [42] bring the potential of near-to-medium-future hepatocyte replacement therapies into sharper focus. As CPS1 deficiency has been treated in limited cases with primary hepatocyte transplantation [19], this approach has untapped therapeutic potential, especially in the context of ongoing donor liver scarcities [43].

In conclusion, these studies have demonstrated that CPS1 patient-derived iPSCs may be developed successfully to generate hepatocyte-like cells; however, EF1α promoter-driven hcoCPS1 expression from the AAVS1 safe harbor is not suitable to reconstitute the urea cycle. This work lays a foundation for future studies investigating alternative promoter constructs and cis-regulatory elements to optimize transgene expression in differentiated cells and restore activity for this critical hepatic enzyme.

Materials and Methods

Cell Lines and Reprogramming

Deidentified CPS1 deficiency patient fibroblasts were purchased from Coriell Institute for Medical Research (GM03432, Camden, NJ) or were a generous gift from Johannes Häberle (CD1 and CD2). The mutations in all lines were confirmed by Sanger sequencing (Laragen, Culver City, CA) and independently verified by exome sequencing performed by Prevention Genetics (Marshfield, WI). Fibroblasts were cultured on 0.1% gelatin (07903, StemCell Technologies, Vancouver, Canada) in DMEM/F12 (11320033) supplemented with 20% fetal bovine serum (FBS; 10437028), 1% glutaMAX (35050061), 1% non-essential amino acids (NEAA; 11140050, all from Invitrogen, Carlsbad, CA), and primocin (ant-pm-1, Invivogen, San Diego, CA). Fibroblasts were reprogrammed using the lentiviral STEMCCA vector expressing the four Yamanaka factors as described previously [26] and were subsequently cultured on irradiated mouse embryonic fibroblasts (A34180, ThermoFisher, Waltham, MA) in mTeSR Plus growth media (100-0276, StemCell Technologies) until colonies emerged. Clones were picked and transduced with AAV8-Cre (a gift from James M. Wilson, Addgene viral prep 105537-AAV8) to remove the STEMCCA cassette. Genomic DNA was harvested (69504, Qiagen, Hilden, Germany), and STEMCCA removal was confirmed by PCR as described (Center for Regenerative Medicine [CReM] iPSC Core of Boston University and Boston Medical Center https://www.bu.edu/dbin/stemcells/index_iPSC.php). Primers for STEMCCA PCR were: Forward: 5’-TGGCTCTCCTCAAGCGTATT-3’; Reverse: 5’-GTTGTGCATCTTGGGGTTCT-3’. iPSCs were routinely maintained on hESC-qualified Matrigel (354277, Corning Inc., Corning, NY) in mTeSR Plus and passaged using ReLeSR (05872, StemCell Technologies) according to the manufacturer’s instructions.

Immunocytochemistry

iPSCs were probed for pluripotency markers as described previously [26]. Primary antibodies used were: OCT3/4 (SC-5279, Santa Cruz Biotechnology, Dallas, TX); KLF4 (RC-09-0021, Stemgent, Cambridge, MA); TRA1-60 (mab4360, MilliporeSigma, Burlington, MA); TRA1-81 (mab4381, MilliporeSigma) all at 1:200 dilutions. Secondary antibody used was goat anti-mouse IgG Alexa Fluor 488 (A-11001, Life Technologies, Carlsbad, CA) at 1:200 dilution.

qPCR

qPCR was performed as previously described [13]. Briefly, total RNA was isolated, and cDNA synthesized according to manufacturer’s instructions (11828665001 and 4897030001, respectively, Roche, Basel, Switzerland). qPCR was carried out using ssoAdvanced Universal SYBRGreen (1725271, BioRad, Hercules, CA). A list of primers used can be found in Supplemental Table 1 (Appendix A). GAPDH was used as the housekeeping gene, and fold changes were calculated using the -ΔΔCt method.

Teratoma Assay

All mice were kept according to the National Institutes of Health guidelines and all experimental procedures were approved by and conducted according to the guidelines of the Institutional Animal Care and Use Committee of the University of California, Los Angeles. Mice had ad lib access to standard chow and water and were maintained on a 12-hour light-dark cycle. Severe combined immunodeficient mice (NOD.Cg-Prkdc Il2rg/SzJ [NSG], Stock Number 005557, Jackson Laboratory, Bar Harbor, ME) were anaesthetized with isoflurane, and 10 million iPSCs suspended in Matrigel were injected unilaterally into the quadriceps. Mice were euthanized when visible tumors formed, and the tissue was fixed in 10% buffered formalin for 48 hours. Teratomas were embedded in paraffin, sectioned, and stained with hematoxylin and eosin for imaging.

Hepatic Differentiation

iPSCs were differentiated to hepatocyte-like cells as described previously [28] with modifications. Undifferentiated iPSCs were incubated in 10µM Y-27632 ROCK inhibitor (SM-008, BioPioneer, San Diego, CA) for 1 hour and subsequently dissociated using Accutase (A1110501, ThermoFisher). Cells were then seeded onto Matrigel (CB-40230C, Corning Inc.) at a density of 2 million per well of a 6-well plate (approximately 2.1x105 cells/cm2). The following day, cells were induced to form endoderm using the STEMDiff Definitive Endoderm kit (5110, StemCell Technologies) for 4 days. Endodermal cells were then replated at ~50% density (empirically optimized for each individual cell line) in differentiation media made with DMEM/F12 containing 10% knockout serum replacement (10828028, Invitrogen), 1% NEAA, 1% glutaMAX, and primocin. This was further supplemented with 1% DMSO (D8418-500ML, MilliporeSigma) and 100ng/mL hepatocyte growth factor (78019, StemCell Technologies) for 8 days. Cells were then cultured for 6 days in differentiation media supplemented with 0.1µM dexamethasone (D4902-25MG, MilliporeSigma). Finally, cells were matured for 3 days in William’s E media (12551032, Invitrogen) containing 10% FBS, 1% NEAA, 1% glutaMAX, 1.8% DMSO, 1µg/mL human insulin (I9278-5ML, MilliporeSigma), 4.8µg/mL hydrocortisone 21-hemisuccinate (H2270-100MG, MilliporeSigma), and 0.1µM dexamethasone. Media was changed daily throughout the differentiation. Cells were harvested using 0.25% trypsin (25200056, Invitrogen).

Molecular Cloning and CRISPR/Cas9 Genome Editing

Human codon-optimized CPS1 (hcoCPS1) was synthesized by Blue Heron Biotech (Bothell, WA) as previously described [13]. hcoCPS1 was cloned into the ARC vector and inserted into the AAVS1 site as described previously [27]. The plasmid ARC containing homology arms to AAVS1, a splice acceptor site, T2A, and puromycin resistance gene was used as the starting backbone, which was digested with KpnI and EcoRV to generate the vector backbone. NEBuilder HiFI DNA Assembly (E2621S, New England Biolabs, Ipswitch, MA) was used to introduce EF1α-driven hcoCPS1, along with the human growth hormone polyadenylation signal, into digested ARC, which was subsequently confirmed by sequencing, replacing the original transgenic construct and forming the new donor. This donor was nucleofected into iPSCs along with paired Cas9 nickases (a gift from Feng Zhang [44], Addgene #48140) targeting AAVS1, using the Amaxa Nucleofector (Lonza, Basel, Switzerland) program B016. Treated cells were then selected with puromycin at 1ug/mL. Clones were isolated and screened by PCR using the primers in Supplemental Table 1 (Appendix A). Faithful transgene integration was confirmed by sequencing (Laragen).

In Vitro Ammonia Challenge

Day 20 HLCs were treated with 2.5mM 15NH4Cl (Cambridge Isotope Laboratories, NLM-467-1), and supernatant was harvested after 24h and stored at -80°C until analysis as described previously [45-47]. Ammonia levels were detected using a colorimetric ammonia assay (ab83360, Abcam, Cambridge, UK) according to the manufacturer’s instructions.

Albumin Secretion

Day 20 HLCs received fresh growth media, which was then harvested from day 21 HLCs after 24 hours. Albumin concentration was measured by ELISA according to the manufacturer’s instructions (E80-129, Bethyl Laboratories, Montgomery, TX).

Urea Secretion

Day 20 HLCs received fresh growth media, which was then harvested from day 21 HLCs after 24 hours. Urea concentration was measured by colorimetric assay according to the manufacturer’s instructions (EIBUN, ThermoFisher).

DNA Methylation Analysis

Genomic DNA from iPSCs and HLCs was isolated (69506, Qiagen) and subsequently bisulfite converted (D5005, Zymo Research, Irvine, CA) according to the manufacturer’s instructions. The bisulfite-converted 231bp core EF1α promoter was PCR amplified and sequenced. Primers were designed using the Zymo Bisulfite Primer Seeker tool and are listed in Supplemental Table 1 (Appendix A).

Statistical Analysis

Collected data was analyzed with the Prism8 (GraphPad, San Diego, CA) statistical package. Results were expressed as mean ± standard error of the mean (SEM) and p values were determined using one-way ANOVA with Tukey’s multiple comparison’s test, or unpaired T-test when applicable. p < 0.05 was considered significant.