Generation and phenotyping of huR83C, a humanized G6PC-R83C mouse model for GSD-Ia
Mammalian G6Pase-α proteins share 87–91% amino acid sequence identity and are functionally equivalent (1–3). We have shown that human and mouse G6Pase-α have identical structure-function relationships, and that the G6pc-R83C variant is pathogenic in mice (15). In this study, we used wild-type mice (mR83) carrying two functional mouse (m) G6pc-R83 alleles, the homozygous (huR83C) mice carrying two alleles with the human (hu) G6PC-R83C variant knocked-in at the G6pc locus, and the heterozygous (mR83/huR83C) mice. GSD-Ia is an autosomal recessive disorder (1–3), and, as expected, the mR83 and mR83/huR83C littermates display indistinguishable wild-type phenotypes and, therefore, were used as controls.
Liver microsomal G6Pase-α activity in 3-week-old wild-type, heterozygous, and homozygous huR83C mice averaged 330.4 ± 36.3, 176.2 ± 13.7, and 2.2 ± 0.3 units, respectively. Kidney microsomal G6Pase-α enzyme activity averaged 607.2 ± 49.2, 267.5 ± 8.2, and 2.6 ± 0.4 units, across respective cohorts. The lower level of quantitation for the microsomal G6Pase-α assay is 2 units, indicating that the huR83C mice have null hepatic and renal G6pc loci. As expected, the huR83C mice manifest a GSD-Ia phenotype of impaired glucose homeostasis (1–3) as shown below in Figs. 2 and 3. Consistent with this, of the n = 42 untreated huR83C mice studied, only 16 (39%) survived to age 3 weeks.
Base editing strategy for correction of G6PC-R83C
Genetic correction of the pathogenic human G6PC-R83C (G6PC-g.247C > T) variant requires the conversion of an A:T base pair to a G:C base pair. To target this allele, a guide RNA was designed with an saCas9 ‘NNGRRT’ protospacer-adjacent motif (PAM), with target adenine (noted at position 16 in Fig. 1a). The base-editor used harbors an saCas9 nickase domain and a TadA deaminase domain engineered to maximize editing of the target adenine and reduce editing of bystander adenines (24). The guide RNA and messenger RNA encoding the base-editor were co-formulated into an LNP to generate BEAM-301, an investigational base-editing therapy for GSD-Ia patients who harbor the G6PC-R83C variant. Editing at the target adenine corrects the g.247C > T/p.R83C variant back to wild-type g.247C/p.R83. There are two neighboring bystander adenine bases at positions 18 and 22 (Fig. 1a). An edit at position 18 introduces a synonymous variant while an edit at position 22 introduces a nonsynonymous p.Y85H variant.
Given the high rate of neonatal lethality in the homozygous huR83C mice, we first validated that BEAM-301 administration can correct the G6PC-R83C (G6PC- g.247C > T) allele in the livers of 10-12-week-old heterozygous (mR83/huR83C) mice (Fig. 1). BEAM-301 was dosed systemically via tail vein injection of 0.1 to 3.0 mg/kg and the base-editing efficiency of the G6PC alleles was analyzed via next-generation sequencing (NGS) in isolated liver slices at 7 days post-dose (Fig. 1b). The editing efficiency for the conversion of the GSD-Ia G6PC-R83C variant to wild-type G6PC-R83 (G6PC-g.247C) correlated positively with administered dose level of BEAM-301. A low frequency of both bystander edits was also detected. The functional impact of the non-synonymous Y85H edit (position 22) was evaluated using transient expression assays in COS1 cells (28). The G6Pase-α-Y85H variant lacked detectable phosphohydrolase activity (Supplementary Fig. 2), classifying it as a pathogenic G6Pase-α variant.
The editing efficiency of BEAM-301, administered to heterozygous mice at 1.5 mg/kg (301H) is shown in Fig. 1c. An average of 39.1% of sequencing reads encoded wild-type G6Pase-α, whereas 5.1% contained the nonsynonymous bystander edit resulting in the Y85H allele, 0.8% contained only a synonymous bystander edit, and 0.7% exhibited indels. Since the predominant editing outcome is correction to wild-type G6PC-R83, and all other outcomes are expected to yield inactive G6Pase-α, similar to the pre-existing G6PC-R83C variant, the editing induced by BEAM-301 is expected to provide an overall functional benefit.
Pathophysiological analysis of 301H-dosed huR83C mice at 3 weeks of age
Newborn (NB) huR83C mice were treated by systemic administration of BEAM-301, at a dose level of 1.5 mg/kg (301H), and metabolic correction was assessed at 3 weeks of age. Of the eight mice treated, two mice expressed background hepatic and renal G6Pase-α activity of 2.0 and 2.2 units, respectively, and genomic analysis confirmed the absence of base-editing in their livers, suggesting failed administration of BEAM-301 rather than failure to target the locus. Therefore, they were excluded from further analysis. In the remaining six mice, liver microsomal G6Pase-α activities averaged 163.8 ± 48.5 units (Fig. 2a), statistically equivalent to heterozygous control mice which had 176.2 ± 13.7 units. Individually, hepatic G6Pase-α activity ranged from 33 to 289 units (Fig. 2a), and revealed a positive correlation with base-editing efficiency (R83C > R83), with a maximum rate of ~ 60% (Fig. 2b). Published data show that restoring ≥ 5 units of hepatic G6Pase-α activity in GSD-Ia mice enables them to maintain blood glucose homeostasis (9–11, 15, 16). As expected, at age 3 weeks and in the fed state, blood glucose levels in 301H-treated huR83C mice were similar to their control littermates (Fig. 2c). Additionally, the 301H-treated huR83C mice were able to survive 24 hours of fasting challenges, with fasting blood glucose levels indistinguishable from the control mice (Fig. 2c). In contrast, the untreated huR83C mice exhibited marked hypoglycemia within 60 to 75 min of fasting (Supplementary Fig. 3), a hallmark of GSD-Ia (1–3) as was previously shown for the untreated G6pc-/- mice (10).
Studies have shown that G6Pase-α activity in wild-type (mR83) or heterozygous (mR83/huR83C) mice is expressed in hepatocytes and kidney cortex (1–3). Consistently, histochemical enzyme analysis showed that in the heterozygous (mR83/huR83C) mouse, hepatic G6Pase-α activity was distributed throughout the hepatocytes, with higher levels in proximity to the blood vessels, while in the mR83/huR83C mouse kidney, G6Pase-α activity was restricted to the cortex (Fig. 2d). As expected, G6Pase-α activity was not detectable in the liver or kidney sections of the huR83C mice (Fig. 2d). In the NB-301H-dosed huR83C mice expressing wild-type levels of hepatic G6Pase-α activity, the distribution pattern was similar to that of the mR83/huR83C control mice (Fig. 2d). The edited mice expressing lower levels of hepatic G6Pase-α activity had a focal pattern of expression with G6Pase-α enzymatic activity associated with the vasculature (Fig. 2d). These findings are consistent with those observed in rAAV-mediated gene augmentation in G6pc-/- mice (10). Consistent with the enzyme activity assays, no histochemical staining of G6Pase-α was detected in the kidneys of the edited huR83C mice (Fig. 2d).
The untreated huR83C mice manifested a typical GSD-Ia phenotype of hepatomegaly, nephromegaly, and growth retardation (Fig. 2e and 2f). At 3 weeks of age, the body weight (BW) of huR83C mice measured significantly lower than their sex-matched control littermates (Fig. 2e and 2f) and their liver weight (LW)/BW and kidney weight (KW)/BW ratios were 13.3 ± 0.8% and 5.2 ± 0.2%, respectively, which were significantly higher than the respective ratios in the control mice of 4.2 ± 0.1% and 1.30 ± 0.02% (Fig. 2e). In the NB-dosed huR83C mice, the BW, LW/BW, and KW/BW values closely approximate those of the control littermates (Fig. 2e). Despite renal G6Pase-α activity being non-detectable in both untreated and NB-dosed huR83C mice at 3 weeks of age, restoration of liver G6Pase-α activity by base-editing benefitted the kidney, with the KW/BW value decreasing from 5.2 ± 0.2% for the untreated huR83C mice to 2.4 ± 0.1% for the edited mice, compared to 1.30 ± 0.02% for the control mice (Fig. 2e). In the 3-week-old 301H-dosed mice, the LW/BW values were inversely correlated with the level of hepatic G6Pase-α activity restored (Fig. 2g), demonstrating a direct relationship between liver G6Pase-α activity and hepatomegaly. Hepatomegaly in GSD-Ia is caused by elevated accumulation of glycogen/neutral fat (1–3). Notably, in the NB-dosed mice, hepatic levels of both glycogen and triglyceride were inversely correlated with the level of restored hepatic G6Pase-α activity (Fig. 2g). In the NB-dosed mice, hepatic G6P levels were also inversely correlated with hepatic G6Pase-α activity restored (Fig. 2g).
The untreated huR83C mice also manifested hypoglycemia, hypertriglyceridemia, hypercholesterolemia, and hyperuricemia (Fig. 3a), characteristic of GSD-Ia (1–3). Serum lactate concentrations were not significantly elevated in 3-week-old, untreated huR83C mice (Fig. 3a), consistent with previous observations in the G6pc-/- mice (21). Compared to untreated huR83C mice, the NB-dosed mice displayed improved blood/serum metabolite profiles (Fig. 3a). A deficiency in G6Pase-α reduces hepatic glucose production and reprograms G6P metabolism, leading to increased G6P accumulation, glycogen synthesis, and glycolysis (29). Compared to the controls, hepatic glucose levels in the 3-week-old untreated huR83C mice were markedly reduced along with increased hepatic levels of triglyceride, glycogen, lactate, and G6P (Fig. 3b). The NB-dosed mice displayed normal hepatic levels of triglyceride, lactate, and glycogen, although hepatic glucose levels remained reduced and hepatic G6P levels remained elevated, compared to the controls (Fig. 3b).
Hematoxylin and eosin (H&E) staining showed that the 3-week-old untreated huR83C mice displayed no histological abnormalities in the liver except for a diffuse mosaic pattern of hepatocytes, consistent with their increased glycogen accumulation (Fig. 3c), as seen in GSD-Ia patients. H&E staining also showed that the 3-week-old untreated huR83C mice displayed marked glycogen accumulation in the renal tubular epithelial cells, resulting in enlargement and compression of the glomeruli (Fig. 3c). Oil Red O staining confirmed the increases in neutral fat accumulation in the liver with little or no fat accumulation in the kidney of 3-week-old huR83C mice (Fig. 3d), consistent with GSD-Ia and the G6pc-R83C mice (15).
Compared to 3-week-old untreated huR83C mice, H&E and Oil Red O staining confirmed the marked reduction in hepatic levels of glycogen and neutral fat in the NB-dosed mice that inversely correlated to the extent of hepatic G6Pase-α activity restored (Fig. 3c & 3d). Little to no fat accumulation was observed in the edited kidneys (Fig. 3d). In summary, the huR83C mice manifested a phenotype of impaired glucose homeostasis that mimics human GSD-Ia, validating a humanized knock-in mouse model for GSD-Ia. Moreover, the NB-dosed huR83C mice displayed a markedly improved metabolic phenotype, establishing the efficacy of base-editing in correcting metabolic abnormalities in GSD-Ia mice.
Pathophysiology of 301H-dosed huR83C mice at 8 weeks of age
In contrast to the tight range of editing rates observed in BEAM-301-treated adult heterozygous (mR83/huR83C) mice (Fig. 1), homozygous (huR83C) mice edited at birth exhibited a large variation in both the editing efficiency and the hepatic G6Pase-α activity restored (Fig. 2).
To further investigate this phenomenon, we infused 301H into NB and 3W huR83C mice and compared their phenotypes at age 8 weeks. Liver microsomal G6Pase-α activity in 8-week-old wild-type and heterozygous mice measured 258.9 ± 12.5 and 174.5 ± 7.0 units on average, respectively (Fig. 4a). At age 8 weeks, liver microsomal G6Pase-α activity in 3W-dosed huR83C mice averaged 150.8 ± 27.4 units, over 2-fold higher than NB-dosed mice that averaged 65.0 ± 22.8 units (Fig. 4a). Liver microsomal G6Pase-α activities in the NB-dosed mice varied 35-fold, from 5 to 173 units, with a corresponding range in base-editing efficiency of 0.8–35% as analyzed via NGS. In contrast, liver microsomal G6Pase-α activities in the 3W-dosed mice varied only 10-fold, from 30 to 294 units, with a range in base-editing efficiency of 2.9–36.6%. Based on these data, dosing huR83C mice with 301H at 3 weeks of age appears to reduce editing variability and increase editing efficiency. Again, restoration of hepatic G6Pase-α activity correlated with the editing efficiency for all cohorts dosed, irrespective of age at dosing (Fig. 4b). Importantly, all NB (n = 8) and 3W (n = 9) 301H-dosed huR83C mice survived to the terminal time-point of the study (8 weeks) and could sustain 24 hours of fasting (Fig. 4c).
In the kidney of huR83C mice, systemic infusion of 301H restored little or no G6Pase-α expression with kidney microsomal G6Pase-α activity in NB-dosed (4.5 ± 0.44 units) and 3W-dosed (3.3 ± 0.51 units) much lower than that in the 8-week-old wild-types (520.8 ± 55.5) and heterozygotes (270.2 ± 18.8 units) (Fig. 4a).
At age 8-weeks, the BW values of 3W-301H-dosed huR83C mice were lower than both their sex-matched littermate controls and NB-301H-dosed huR83C mice (Fig. 4d), which can be explained by the significantly lower baseline BW values of the 3-week-old huR83C mice prior to dosing (Fig. 2f). The 8-week-old 301H-dosed huR83C mice continued to display hepatomegaly and nephromegaly (Fig. 4d) and the LW/BW values and hepatic levels of glycogen and G6P were inversely correlated with the level of hepatic G6Pase-α activity restored (Fig. 4e).
Importantly, all 301H-dosed huR83C mice displayed normal blood/serum levels of glucose, cholesterol, triglyceride, lactic acid, and uric acid (Fig. 4f). Hepatic levels of triglyceride in all edited huR83C mice were similar to those of littermate controls but hepatic levels of glycogen and G6P were elevated (Fig. 4g). Notably, at age 8 weeks, hepatic levels of glucose and lactate were completely normalized in the 3W-dosed but not in the NB-dosed huR83C mice (Fig. 4g).
Pathophysiology of 301L-dosed huR83C mice at 8 weeks of age
The efficacy of reducing the dose of editing reagents was assessed by repeating the earlier experiment at half the original dose of BEAM-301, namely 0.75 mg/kg (301L) and monitoring the restoration of metabolic parameters at 8 weeks of age. Among the 18 huR83C mice treated as newborns, 9 died prematurely, resulting in an 8-week survival rate of 50%. Among the 21 huR83C mice treated at age 3 weeks, 3 died prematurely, resulting in an 8-week survival rate of 86%. Due to the broad range in hepatic G6Pase-α restored in NB-dosed mice noted above, it is likely that 50% of the low-dose, NB-treated huR83C mice failed to express sufficient hepatic G6Pase-α activity required for long-term survival.
The liver microsomal G6Pase-α activity in 8-week-old control mice, the surviving NB-dosed (n = 9) and 3W-dosed (n = 8) huR83C mice averaged 238.9 ± 17.6, 24.5 ± 5.1 and 31.3 ± 7.3 units, respectively (Fig. 5a). Again, hepatic G6Pase-α activity restored in the edited mice correlated with editing efficiency (Fig. 5b). Moreover, all 8-week-old edited huR83C mice could sustain 24 hours of fasting (Fig. 5c).
As was seen with the high dose, the low dose (301L) 3W-dosed huR83C mice had significantly lower BW values than their littermate controls and were also lower than the NB-treated huR83C mice, most likely due to the lower baseline weight prior to dosing at 3W (Fig. 5d). Both the NB and 3W 301L-dosed cohorts continued to manifest hepatomegaly and nephromegaly (Fig. 5d). The LW/BW values and hepatic levels of glycogen and G6P of the 8-week-old 301L-dosed mice were inversely correlated with the level of hepatic G6Pase-α activity restored (Fig. 5e), as expected by the direct relationship between liver G6Pase-α activity and metabolic control. While the NB-dosed huR83C mice displayed increased levels of serum cholesterol (Fig. 5f), the 3W-dosed mice displayed a normal blood/serum metabolite profile at age 8 weeks (Fig. 5f), reflecting the lower hepatic G6Pase-α activity restored in the NB-dosed mice compared to the 3W-dosed mice. Compared to the 8-week-old littermate controls, both NB- and 3W-dosed huR83C mice displayed normal hepatic triglyceride levels, although their hepatic levels of glucose remained reduced and hepatic levels of glycogen, G6P, and lactate remained elevated (Fig. 5g).
Long-term correction of metabolic abnormalities of the huR83C mice
One severe long-term complication of GSD-Ia is HCA which may undergo malignant transformation to HCC (1–3, 8). To evaluate the long-term efficacy of base editing, we undertook a study up to 53 weeks of age by infusing NB huR83C mice with a high dose (301H) and 3W huR83C mice with the lower half-dose (301L) of the BEAM-301 editing reagents.
There was no premature death of the NB-301H-edited huR83C mice. Liver microsomal G6Pase-α activity in 53-week-old wild-type and NB-dosed huR83C mice averaged 210.6 ± 5.9 and 87.1 ± 19.9 units, respectively (Fig. 6a), demonstrating sustained hepatic G6Pase-α expression. Analysis of liver isolates from 53-week-old NB-dosed huR83C mice revealed base-editing rates ranging from ~ 0.5–53% and hepatic G6Pase-α activity from 2.4 to 292 units. Consistent with assessment at earlier terminal time-points, hepatic G6Pase-α activity restored in the NB-dosed mice correlated positively with base-editing efficiency (Fig. 6b). At 53 weeks of age, all NB-dosed huR83C mice (n = 19) could sustain 24 hours of fasting (Fig. 6c).
At age 53 weeks, the BW and LW values of wild-type and NB-301H-dosed huR83C mice were statistically similar but the NB-dosed mice continued manifesting hepatomegaly and nephromegaly (Fig. 6d). In congruence with earlier data from younger mice, the LW/BW values, and hepatic levels of glycogen and G6P were inversely correlated with the level of hepatic G6Pase-α activity restored (Fig. 6e). The 53-week-old NB-dosed huR83C mice displayed normal levels of blood/serum metabolites, except for moderately elevated serum triglyceride levels (Fig. 6f). Compared to the wild-type controls, these NB-dosed huR83C mice displayed normal levels of hepatic glycogen, triglyceride, and lactate, although their average hepatic glucose levels remained reduced and average hepatic G6P levels remained elevated (Fig. 6g).
Among the 21 huR83C mice treated with 301L at age 3 weeks, 3 mice died before age 8 weeks, an additional 8 mice underwent scheduled euthanasia at 8 weeks of age, and 1 more died at age 39 weeks, resulting in a 53-week survival rate of 69%. Liver microsomal G6Pase-α activity in 53-week-old control (mR83/huR83C; n = 14) and 3W-301L-dosed huR83C (n = 9) mice averaged 180.1 ± 13.5 and 53.4 ± 17.6 units, respectively (Fig. 7a). The range of hepatic G6Pase-α activity in the 301L-dosed huR83C mice was 2.6 to 179 units, with positive correlation to base-editing, that ranged from a rate of 0.12–36% (Fig. 7b). Except for the dosed huR83C mouse expressing 2.6 units of hepatic G6Pase-α activity, the other 8 edited mice could sustain 24 hours of fasting (Fig. 7c).
At age 53-weeks, the BW values of 3W-301L-dosed huR83C mice were significantly lower than their age-matched control mice (Fig. 7d). Compared to 53-week-old control mice, the 3W-301L-dosed huR83C mice continued manifesting hepatomegaly and nephromegaly (Fig. 7d), although their LW values were statistically similar to the control mice. Again, the LW/BW and hepatic levels of glycogen and G6P were inversely correlated with hepatic G6Pase-α activity restored (Fig. 7e).
Compared to the controls, the 53-week-old 3W-301L-dosed huR83C mice displayed normal levels of blood/serum metabolites (Fig. 7f) and normal levels of hepatic glycogen and lactate, although their hepatic glucose levels remained reduced and their hepatic G6P levels remained elevated (Fig. 7g). Interestingly, hepatic triglyceride levels in the 53-week-old 301L-dosed huR83C mice were markedly lower than their sex-matched littermate controls (Fig. 7g).
We have previously shown that the rAAV-G6PC-treated G6pc-/- mice are leaner and protected against age-related insulin resistance (30). At age 53 weeks, the BMI values of the 3W-301L-dosed huR83C mice were significantly lower than that of the sex-matched littermate controls (Fig. 8a), confirming that the edited mice were leaner. For insulin tolerance test, a reduced insulin dose of 0.25 IU/kg was used because GSD-Ia mice have an increased insulin sensitivity (30). Following an intraperitoneal insulin injection, blood glucose levels in the 53-week-old control mice failed to decrease (Fig. 8b), reflecting an age-related decrease in insulin sensitivity (30, 31). Conversely, following an intraperitoneal insulin injection, blood glucose levels in the 53-week-old 3W-301L-dosed huR83C mice decreased with time (Fig. 8b), demonstrating that the edited mice were protected against an age-related insulin resistance, consistent with previous observations in the rAAV-treated G6pc-/- mice (30).
Enzyme histochemical analysis showed that enzymatically active G6Pase-α in 53-week-old control mice was distributed throughout the liver with significantly higher levels in proximity to blood vessels (Fig. 8c), similar to that observed at 3 weeks of age. In the 53-week-old 3W-301L-dosed huR83C mice, G6Pase-α was also distributed throughout the liver, although less uniformly, with foci containing markedly higher levels of enzymatic activity and other regions harboring little or no G6Pase-α activity (Fig. 8c).
H&E staining showed that both the 53-week-old control and 3W-301L-dosed huR83C mice exhibited no hepatic histological abnormalities (Fig. 8d). Oil red O staining showed that the 53-week-old control mice showed a variable degree of neutral fat storage (Fig. 8d). Among the six control mice analyzed, four displayed high levels of neutral fat storage. In contrast, among the six 3W-301L-dosed huR83C mice, only the edited mouse expressing 2.6 units of hepatic G6Pase-α activity showed high levels of neutral fat storage (Fig. 8d), while the other five edited mice showed little or no hepatic fat storage. This is consistent with the observation that hepatic triglyceride levels in the 3W-301L-53W mice were significantly lower than their littermate controls (Fig. 7g). In summary, at age 53 weeks, all BEAM-301-dosed huR83C mice displayed a near normal metabolic phenotype and were leaner and protected against age-related insulin resistance.