Crigler-Najjar Syndrome

Crigler-Najjar syndrome (CNS) is a recessive inherited disorder caused by a deficiency of uridine diphosphoglucuronosyl transferase 1A1 (UGT1A1). Because of missing hepatic bilirubin glucuronidation, patients experience accumulation of unconjugated bilirubin in serum, leading to a risk of irreversible damage to the central nervous system. For CNS, a natural mutant rat model exists, the Gunn rat, and also a transgenic mouse model was described. The mouse model requires phototherapy for survival to adulthood, whether or not it’s the Gunn rat. The CNS is an ideal model for GT because liver parenchyma in classical CNS is considered structurally and histologically normal; however, it is reported that over time, some fibrosis can develop. Further, the correction level needed for sufficient phenotypic correction is low—only 5% of normal UGT1A1 activity. Finally, the success of the intervention is easily monitored by serum bilirubin levels. Up to now, the only available conservative therapy to individuals affected by the severe form of CNS is phototherapy, which has major implications on the quality of life and loses efficacy during adolescence. LT during adolescence is often considered to be the only curative option available to prevent acute decompensation and irreversible brain damage. For CNS, various GT approaches have been tested in animal models. The rAAV approach seems to be the most promising. Seppen et al. showed a 70% reduction of serum bilirubin levels following AAV gene transfer for UGT1A1. The dose used in this study (2.5–5 × 10 ¹² vector genomes [vg]/kg body weight) was similar to that used in the high-dose cohort of the rAAV8 trial of GT for hemophilia B, which showed efficacy, although with detection of immune responses against the viral capsid in some patients. In the UGT1A1 knockout mouse model, Bortolussi et al. were able to show phenotypic correction after intraperitoneal injection of AAV9 UGT1A1 vectors in neonate mice.

More recent work in Gunn rats shows that AAV8-mediated gene transfer to the liver using an optimized UGT1A1 expression cassette results in correction of bilirubin levels at doses lower than those tested so far, and also, studies in the mouse model showed sufficient therapy with 2.5 × 10 ¹¹ vg/kg body weight.

Like in many LBMDs, it would be suitable to treat the disease in children, first, to eliminate the risk for neurological damages, second, to avoid the quality of life impairing phototherapy as early as possible, and third, to avoid development of liver fibrosis (based on very recent findings ). But this is difficult because of the described problem of transgene-loss in rAAV8 episomal GT in the growing liver.

Two recent studies tested the readministration in the Crigler-Najjar mouse model with success. Greig et al. readministered the same vector on day 56 after first injection within the first 24 hours of life. Because of the very early first injection, mice did not develop antibodies, which is also described by Wang et al. However, this approach is very unlikely to be transferred to the clinic because such an early diagnosis in the newborn is impossible except when genetic testing is performed in pregnancy because of known risk. More translational is the approach of Bockor et al., with serotype switching for readministration. For this approach, the need for the approval of two products could hamper clinical translation. Better solutions could be to find possibilities for readministration or targeted integration (see “AAV Persistence in the Developing Liver”). One very attractive targeted integration approach was studied in the Crigler-Najjar mouse model by Porro and colleagues. They could reduce bilirubin levels in the mice by intraperitoneal injection of rAAV2/8 packed with a template DNA consisting of homologous arms for the albumin locus so that UGT1A1 cDNA is integrated into the host genome at the end of the albumin reading frame separated from the albumin by a 2a peptide. This approach reached 5% enzyme activity of the wild type, enough for significant reduction of serum bilirubin levels, without any nuclease-enhancing homology-directed repair/homologous recombination (HDR) but with high vector doses (1 × 10 ¹² vg/mouse). Therefore for clinical translation, this approach needs to be optimized.

Glycogen Storage Disease Type 1

Glycogen storage disease type 1a (GSD1a) is caused by glucose 6-phosphatase (G6Pase) deficiency, which is expressed mainly in the liver, the intestines, and the kidney. Long-term complications of GSD1a include life-threatening hypoglycemia, failure to thrive, renal failure, osteoporosis, and hepatic adenomas and carcinomas. Patients with GSD1b additionally suffer from neutropenia and myeloid dysfunction. Dietary therapy has significant limitations, and patients are constantly exposed to the risk of life-threatening hypoglycemia. In some metabolic diseases, protein/enzyme replacement is a treatment option, but this option is expensive, needs high-frequency readmissions, and can be hampered by antibody production. For GSD1, this is no option because the defect G6Pase is an extremely hydrophobic transmembrane protein that is difficult to purify. Different animal studies for GT in GSD1a have been performed with different results. Most show a good metabolic correction and assume prevention of hepatocellular carcinoma (HCC), but others show a decline of transgene expression over time and still a risk for HCC. In the following, some of these studies are summarized.

Luo et al. showed in their study that the mouse model can reach metabolic correction after administration of a double-stranded rAAV vector encoding human G6Pase. Another study showed a metabolic correction following rAAV8 injection in the retro-orbital sinus by the age of 2 to 4 weeks (5 × 10 ¹² to 3 × 10 ¹³ vp/kg) for 90 weeks and, most importantly, showed that treated mice do not develop HCC, whereas all of the nontreated mice died. Treated mice showed a G6Pase activity of between 3% and 128% of normal 90 weeks after gene transfer. This study identified 3% G6Pase-alpha activity compared with the wild type as the threshold for phenotypic correction. Another study identified 2% activity as the threshold for HCC prevention in GSD1a, and Chou et al. identified 6% for GSD1b. But Brooks et al. showed that the majority of GSD1a-diseased dogs (four out of five) develop HCCs despite GT after long-term (4–8 years) follow-up, even though G6Pase activity was 21% to 52%.

Koeberl showed that transgene expression declines over time after successful AAV-mediated GT in young mice and dogs, which represents an additional challenge to the development of GTs for GSD1a. Landau et al. showed that this problem could be solved with the “targeted integration into a safe harbor under help of zinc finger nucleases” approach.

Another unsolved problem for clinical translation of GT for GSD1a is ongoing renal disease after liver-directed GT and for GSD1b, and also ongoing myeloid disease. Different studies show that rAAV8-based GT is not able to cure the renal disease of GSD1a, most likely because of the liver tropism of rAAV8. Some studies show that a switch to rAAV9 serotype could maybe improve kidney transduction, especially when delivered into the renal vein.

In summary, GSD1 is still a very challenging disease for GT because of the high risk of HCC in uncorrected hepatocytes and the different organs that have to be treated, especially kidneys and the myeloid system.

Primary Hyperoxaluria

Primary hyperoxaluria type 1 (PH1) is a rare inborn metabolic disease (1:100,000) characterized by alanine-glyoxylate aminotransferase deficiency, resulting in oxalate overproduction and oxaluria with the development of renal failure. The end stage of the disease is a condition called systemic oxalosis, which is life threatening. At present, curative therapy is preemptive LT or combined liver and kidney transplantation. PH1 is a very attractive disease for GT because the unmet medical need is high, but at the same time it is a challenging disease because treatment is necessary in small children, and transduction efficiency needs to be higher as in CNS, for example. The reason for this is ongoing oxalate overproduction from uncorrected hepatocytes, which is not compensated by corrected hepatocytes in this disease (cell-autonomous behavior). From clinical studies, it is known that replacement of one-third of the liver do not prevent kidney failure secondary to oxaluria. Salido et al. tested an rAAV5- and an rAAV8-mediated GT approach in a mouse model of this disease. Vectors were administered by single-tail vein injection with 5 × 10 ¹¹ to 1.5 × 10 ¹³ vg/kg in 12- to 16-week-old mice. During a short follow-up of 50 days, the hepatic alanine-glyoxylate aminotransferase enzyme activity was 24.9% to 10.6%. A high dose of AAV8 and AAV5 results in oxaluria with no significant difference compared with the wild type, although AAV8 was superior to AAV5. Castello et al. showed, in their study, a successful phenotypic correction with a helper-dependent adenoviral vector approach. This study was performed in mice of similar age, like in Salidos’s study, but with a longer follow-up period (24 weeks). These studies show proof of concept, but major questions for clinical translation were not answered. First, especially in this disease, treatment early in life is necessary (renal damage caused is irreversible), and therefore, episomal rAAV GT is not suitable unless readministration becomes possible. Second, crucial transduction efficiencies have to be reached in humans; in these studies, 40% to 80% of hepatocytes have been transduced, but it is known that transduction efficiency of rAAV is lower in humans than in mice. Third, longer follow-up periods would be necessary to prove long-term persistence. This long-term persistence could be reached with the same approaches as mentioned for CNS, for example, preventing antibodies by rapamycin synthetic vaccine particles (SVP) to enable readmission. This approach could enhance transduction efficiency at the same time because transduction of rAAV is mainly driven by autophagy. Another approach described by Meliani et al. is to envelop rAAV vectors in exosomes. These exo-AAVs are less susceptible to antibody-mediated neutralization, and they enhance transduction efficacy, in this study from 20% to 40%. Even though targeted integration could be a solution for avoiding loss of transgene in the growing liver, this approach will have a low transduction efficacy for curing PH1 because corrected hepatocytes have no selection advantage and the disease behaves cell autonomous. That means that corrected cells can not prevent or reduce production of toxic metabolites from uncorrected cells. For all these new approaches, no studies with PH1 animal models exist so far.

Progressive Familial Intrahepatic Cholestasis

The disease group of progressive familial intrahepatic cholestasis (PFIC) is a group of disorders with intrahepatic cholestasis caused by different inherited defects. Except for PFIC3 (MDR3 defect), all of them have a normal gamma-glutamyl-transferase despite cholestasis. PFIC is caused by a defect of FIC1, mainly expressed in the liver and intestines. PFIC2 is caused by dysfunctional or missing bile salt export protein located only in the liver. In the last years, PFIC4, -5, and -6 caused by defects of tight junction protein 2, farnesoid X receptor, and Myosin VB (MYO5B) have been detected. So far, GT studies for these diseases are rare. Siew et al. presented a study about a hybrid AAV piggyBac transposon vector for GT in neonatal multidrug receptor 3–deficient mice at the European Society for Paediatric Gastroenterology Hepatology and Nutrition (ESPGHAN) Annual General Meeting 2014. They used a piggyBac transposon, on one hand, to overcome the loss of episomal transgene cDNA and, on the other hand, to overcome the low transduction efficiency of transposons. With a vector dose of 5 × 10 ¹¹ vg, they were able to prevent the development of liver fibrosis. For PFIC2, -3, and -4, GT will always have to deal with the remaining risk of HCC in uncorrected hepatocytes. PFIC1 has a significant unmet need, because LT in these patients often causes severe problems attributed to the remaining uncorrected defect of FIC1 outside the liver, especially in the intestine. But on the other hand, this implicates that GT would have to transduce these organs also, which is challenging.

Alpha-1-Antitrypsin Deficiency

Alpha-1-antitrypsin (AAT) deficiency is an inherited disease with a defect location in the liver. The defect AAT causes liver and lung disease. Protein misfolding leads to accumulation of AAT in the endoplasmic reticulum of hepatocytes and causes chronic liver injury, liver fibrosis, and HCC, and the aspect of missing AAT in circulation leads to lung emphysema because of missing antiprotease function. This is the reason why GT for this disease is challenging. One needs either a very high rate of gene correction to ameliorate the liver disease, which still has a remaining risk for HCC; or one needs a dual therapy: a transgene producing AAT for the lung and inhibition of mutant AAT expression in the liver. Li et al. successfully studied such an approach in the mouse model. Further targeted integration approaches are attractive for this disease because corrected hepatocytes will have a selection advantage and will, over time, replace the majority of uncorrected cells, but there is still a remaining risk for HCC.

Song et al. showed a proof of concept for CRISPR- and rAAV-mediated AAT correction, but they were able show only partially restored AAT serum levels and did not evaluate still-ongoing liver disease.

Some studies, like the one from Bjursell et al., focused on the disruption of the mutated gene to cure liver disease. Bjursell et al. used an adenovirus plus CRISP/Cas9 approach for this.

Multiple AAT deficiency studies in patients were conducted. These studies used an AAT transgene with a rAAV1 vector for intramuscular application. The study from Mueller et al. could only achieve 2.5% to 3% of the purported therapeutic level, but this could be shown to be stable over 5 years with some evidence that this low expression could have a clinical impact already. Of course, these approaches do not influence liver disease and risk of transgene antibody development because transgene is not produced in the liver. In contrast, clinical phase I studies are running with small interfering RNA to block the mutant AAT, which could cure the liver disease but has no influence on the pulmonary disease.

Ornithine Transcarbamylase Deficiency

Ornithine transcarbamylase deficiency (OTC) is the most common inherited disorder of urea cycle disorders (UCDs) in humans and causes hyperammonemia. The OTC gene is X chromosome linked and expresses the enzyme activity predominantly in the liver. In addition, heterozygous females may experience life-threatening elevations of ammonia in the blood and brain, leading to irreversible cognitive impairment and death. Available treatment focuses on carefully monitored, lifelong dietary protein restriction combined with alternate pathway therapy using sodium phenylbutyrate. So far, correction of the enzyme deficiency is only possible with LT.

From the beginning of GT, OTC has been one target disease. The first phenotypic correction by adenovirus-mediated gene transfer in the OTC mouse model (spf and spf ^ash ) was described in 1996. The following clinical study with adenovirus ended fatally for the treated patient and was an enormous setback for GT.

In 2006, an OTC animal study from Bell et al. raised concerns about rAAV and insertional mutagenesis. But at this point, it is important to mention that the mouse model used is known for higher risk for liver tumors and that only the rAAV LacZ vector–treated mice and not the rAAV OTC–treated ones have had a higher proportion of liver tumors. In 2012 to 2013, Alexander and Cunningham et al. showed in the adult spf ^ash mouse that after intraperitoneal injection of AAV8 vector encoding liver-specific promoter and murine OTC cDNA, 60% OTC activity was established. Cunningham et al., however, showed that this approach is not sufficient to maintain stable transgene expression for preventing hyperammonemia under short hairpin RNA exposure into adulthood when mice were injected as newborns. Because of a wide spectrum of clinical phenotypes in OTC with high mortality and morbidity in the first 6 months of life, LT may be indicated in infancy. Therefore, a potential alternative therapy should also be applicable in early life.

A recent study from Wang et al. showed sufficient GT with self-complementary rAAV8 in 2-month-old mice (heterozygote OTC mice, heterozygote female mouse model) until 18 months. Wang et al. performed a similar study in the classical OTC mouse model, the spf ^ash mouse. They injected adult mouse with their codon-optimized self-complementary rAAV2/8 vector of 3 × 10 ¹¹ and 1 × 10 ¹¹ genome copies and followed them for 56 days. All showed metabolic correction for this time; the high-dose group showed 68% enzyme activity compared with the wild type at day 56 after injection.

Another very important study in the field of GT was conducted on the spf ^ash mouse by Yang et al. in 2016. They could show a remarkable success for the Cas9- and rAAV8-mediated gene repair approach in newborn spf ^ash mice with a follow-up of 8 weeks and a high-protein diet in week 7. They found an HDR rate of 10% and an OTC activity of 20%. Very surprising, and maybe also important for other diseases, was the fact that the same therapy in 8- to 10-week-old mice resulted in aggravation of the disease because of large deletions on the OTC gene, which erased residual OTC activity in diseased mice. The number of correct HDRs was very low in these mice (0.3%–1.7%). This is a very important safety issue, which has to be taken into account in other GT trials with nucleases.

A further important aspect for GT in UCDs is the zoning of metabolic pathways within the hepatocyte because this can influence the efficiency of GT. For urea cycle defects, even a high number of transgene expressions can have low reduction of toxic products in animal studies, because rAAV8 produces transgene expression predominantly in pericentral hepatocytes in mice and dogs, but the urea cycle pathway takes place mostly in periportal hepatocytes. However, in nonhuman primates, rAAV8 shows transgene expression mostly in periportal hepatocytes. Therefore in diseases with metabolic zonation in the liver (for example, UCDs), it is difficult to extrapolate vector doses for human translation from studies in mice.

Tyrosinemia Type 1

Tyrosinemia type 1 (HT1) is an autosomal recessive disease based on fumarylacetoacetate hydrolase leading to abnormal tyrosine degradation and the accumulation of toxic metabolites. Although the tyrosinemia mouse model is a suitable model for GT studies because the phenotype is clear and corrected hepatocytes have a selection advantage, it is not a suitable disease for translational clinical studies because medical management with nitisinone [2-(2-nitro-4-fluoromethylbenzoyl)-1,3-cyclohexanedione] has revolutionized the prognosis for affected patients. Treatment cannot completely abolish the risk of hepatocarcinogenesis; however, this is also a main obstacle for GT in tyrosinemia, because uncorrected cells have a high risk of developing HCCs.

Paulk et al. showed that an AAV-mediated gene correction approach is feasible in vivo and can functionally correct the disease in both adults and neonatal mice with a point mutation on the Fah gene. rAAV8- Fah vectors with homologous sequences for HDR were injected into 3-day-old and 8- to 12-week-old mice intravenously at 1 to 2 × 10 ¹¹ vg/mouse. This is an important result because it showed that gene correction with HDR is feasible also without nucleases. The study by Junge et al. went one step further. They showed that even gene addition via HDR into the ROSA26 locus for the same disease but in another mouse model (FAH ^Δexon5 ) is sufficient to create long-term phenotypic correction without nucleases. However, both studies benefit from the fact that in HT1, corrected hepatocytes have an enormous selection advantage, but this is not the case in many other LBMDs, and further, this also implies that residual uncorrected cells have a high risk of developing HCCs. Therefore translation from this study to clinical studies is questionable. More animal studies on HT1 exist without adding significant new aspects or progress to translational aspects. For example, Shao et al. showed gene repair with adenovirus and Cas9 nickase in a rat model, and Pankowicz et al. applied an approach of metabolic reprogramming by blocking under the help of CRISPR/Cas9, an enzyme one step before fumarylacetoacetate hydrolase to render the phenotype benign. Furthermore, Hickey et al. (in pigs with lentiviral vector) and Van Lith (in mice with rAAV and CRISPR/Cas9) performed studies with successful ex vivo GT.

Wilson Disease

Wilson disease (WD) has a low unmet need for GT because effective medical treatment is available, and in patients with difficult neurological disease or liver cirrhosis at diagnosis, the GT would come too late, like conventional treatment. Otherwise, WD is a suitable disease for GT because corrected hepatocytes have selection advantages, but the risk for HCC in residual uncorrected hepatocytes is not as high as in tyrosinemia, for example, and treatment must not be done in first years of life. One assumes that around 40% of corrected hepatocytes are necessary to correct the phenotype.

There is only one recent study on GT in WD. Murillo et al. showed in a mouse model (6 weeks of age) that rAAV8 (1.5 × 10 ¹² vg/kg) gene addition can reach transduction efficiency, which can cure the phenotype of WD. Further analysis with this approach could even show that GT is capable of reducing the overall cerebral copper content.