Gene modification therapies for hereditary diseases in the fetus

Abstract Proof‐of‐principle disease models have demonstrated the feasibility of an intrauterine gene modification therapy (in utero gene therapy (IUGT)) approach to hereditary diseases as diverse as coagulation disorders, haemoglobinopathies, neurogenetic disorders, congenital metabolic, and pulmonary diseases. Gene addition, which requires the delivery of an integrating or episomal transgene to the target cell nucleus to be transcribed, and gene editing, where the mutation is corrected within the gene of origin, have both been used successfully to increase normal protein production in a bid to reverse or arrest pathology in utero. While most experimental models have employed lentiviral, adenoviral, and adeno‐associated viral vectors engineered to efficiently enter target cells, newer models have also demonstrated the applicability of non‐viral lipid nanoparticles. Amelioration of pathology is dependent primarily on achieving sustained therapeutic transgene expression, silencing of transgene expression, production of neutralising antibodies, the dilutional effect of the recipient's growth on the mass of transduced cells, and the degree of pre‐existing cellular damage. Safety assessment of any IUGT strategy will require long‐term postnatal surveillance of both the fetal recipient and the maternal bystander for cell and genome toxicity, oncogenic potential, immune‐responsiveness, and germline mutation. In this review, we discuss advances in the field and the push toward clinical translation of IUGT.


| INTRODUCTION
The Earth's population, currently at 8 billion, is anticipated to increase to 11 billion by the year 2100 despite decelerating population growth over the past decade. 1 Birth defects incidence will mirror this population growth, particularly in low-and middle-income nations where access to carrier or prenatal screening, genetic diagnostics, or pregnancy interventions is limited, and global migration will increase genetic disease prevalence across high-income nations as well. 2 While the incidence of malformations, as well as chromosomal and autosomal dominant anomalies, is predicted to remain stable, the incidence of autosomal recessive diseases will increase as more rare diseases are recognised as genetic syndromes, 3 [3][4][5][6] More than 400,000 deaths per year are due to birth defects. 4 Computational modeling and clinical trial data for genetic diseases, such as severe combined immunodeficiency syndrome (SCID), transfusion-dependent β-thalassaemia, haemophilia, and spinal muscular atrophy (SMA), demonstrate the substantial health and cost advantages of instituting a cellular or gene modification therapy (GMT) at a younger age and earlier stage of disease [7][8][9][10] ; even greater benefits are expected when the affected fetus is diagnosed and treated in utero.
Gene modification therapy encompasses gene editing and gene addition therapy, both modalities of which are in clinical trials for a range of monogenic disorders, and participants may be as young as a few months of age. 11 A number of these are ideal candidate diseases for in utero therapy due to the detrimental effects of specific protein deficiencies during the critical developmental period. The putative clinical advantages of in utero genetic modification to the affected fetus have been described in numerous reviews. [12][13][14][15] In essence, greater therapeutic efficacy is possible in the fetus compared to the postnatal recipient, given the greater vector-to-mass ratio achieved, the fetal immune system being more tolerant to the administered vector, accessibility through the blood-brain-barrier, and the higher concentration of stem cell targets available, increasing the likelihood of reversing or preventing organ pathology when intervening early in the disease timeline. With vector tolerance achieved by in utero administration, it may be possible to repeat this therapy postnatally to boost levels of the desired protein without immune conditioning. 16,17 In contrast, postnatal therapeutic efficacy has limited impact due to the same factors described, with higher costs for the larger doses required and increased likelihood of a neutralising immune response with age. 18

| In vivo and ex vivo administration
The aim of primary in utero GMT is to perform a single prenatal intervention to correct the aberrant genetic function for the life course of the recipient. 15 This can be achieved with systemic or local administration of the gene therapy tool, via intravenous, intraperitoneal, intraventricular or intracerebral, intratracheal or intraamniotic routes. 22 Gene delivery strategies are designed to introduce the correct DNA sequence (transgene) to the target cell via integration into the host genome or episomally (outside of the host genome) or to introduce a nuclease that can edit the aberrant gene in situ according to a DNA or RNA template in order to revert it to the wild-type form. 23 The gene therapy vehicle can either target the affected organ, such as motor neurons in SMA or Gaucher's disease, or the primary production site of the desired protein, such as the liver for haemophilia. In proof-of-principle animal models, fetal gene therapy has proved more successful than postnatal therapy from the perspective of therapeutic efficacy, reversal of pathology, and improvement of clinical outcomes due to the limited degree of tissue damage earlier in the disease timeline, allowing these organs to produce transgenic proteins and to reverse cellular damage more efficiently. 19,21,24 Younger and less physiologically mature recipient organs are more efficiently transduced and produce greater quantities of transgenic proteins. Further, younger patients have higher disease-free survival as they have less organ pathology, and undamaged target organs are more receptive to transduction. 25 This approach offers a valuable alternative option to pregnancy termination or palliative treatment, and fetal therapy can be performed safely without increasing pregnancy loss. 26 MATTAR ET AL. -675 In utero GMT requires delivery of engineered DNA encoding the desired protein (transgene) packaged in a delivery vehicle (nanoparticle or vector, often of viral origin) designed to display tropism for certain cell types, for example, hepatocytes. The target gene is added to the genome or the host genome is genetically modified, so that the new transgenic protein can be produced by the cell ( Figure 1A). This can be performed ex vivo or in vivo, the former requiring the harvest of stem cells from the recipient, genetic manipulation in vitro and autologous transplantation of modified stem cells, the method employed in clinical trials of gene addition or editing for haemoglobinopathies and hereditary immunodeficiencies. 27 Limited success has already been demonstrated in the context of autologous amniotic fluid-derived haemopoietic cell transplantation in a sheep model. 28 Potential advantages of ex vivo approaches are precise correction of target stem cells without affecting non-target cell types, screening for off-target mutations in transduced cells, and the ability to expand harvested stem cells after gene modification to increase the yield of corrected cells for transplantation. The downsides are greater material costs, longer preparation time, immune-conditioning prior to transplantation and the greater risk to the fetus considering that at least two invasive procedures are required. 29,30 Thus, the in vivo approach would be preferable in utero as this minimises invasive procedures and reduces miscarriage risk. This method is employed in clinical trials of postnatal gene addition therapies using non-integrating viral vectors targeting the haemophilias and SMA 11 and has been used effectively to limit pathology in numerous animal disease models of in utero gene therapy (IUGT).
Advantages include reduced procedural and material costs and direct delivery of vectors to the target organ, while disadvantages include widespread biodistribution to non-target organs in the recipient and mother (due to transplacental spread). The in vivo method has been tested in experimental IUGT models. Potential adverse effects of GMT include progressive loss of transgenic protein expression from transduced cells (diluted over time with the recipient's growth), 21 genotoxicity, 24 immunotoxicity initiated by the vector or transgene, organotoxicity in target or non-target organs, and reproductive toxicity with the possibility of gamete transduction. 17

| Gene addition versus gene editing strategies
Gene addition and gene editing are both applicable to IUGT and have been used to target the same diseases ( Figure 1A or single-strand DNA nicks (base editors), which are repaired by HDR, though non-homologous end joining, which tends to be errorprone, often produces heterogeneous insertions and deletions (indels), and acts concurrently with HDR, reducing editing efficacy. 31

| Delivery tools
In vivo delivery is achieved with viral or non-viral vectors with viral vectors being favored in in utero models because of their superior efficacy. 33,34 Viral vectors derived from adenoviruses (AdV) and adeno-associated viruses (AAV) are preferred as gene addition is episomal with a minimal risk of genome integration. Typically regarded as non-pathogenic vectors, AAV and AdV are associated with immunotoxic reactions which may neutralise transgenic expression, thus diminishing their therapeutic effect, and on the rare occasion this has proven fatal. 35 Typical adverse outcomes with integrating vectors (onco-retroviral, gamma-retroviral, and lentiviral (LV) vectors) include genomic mutagenesis related to obligatory insertion of the transgene into the host genome, which may result in leukemia and other genotoxic events. 24 Non-integrating LV vectors are currently in development to overcome this limitation. 36,37 AdV and integrating vectors are popular vehicles as they can be packaged with large transgenes, while the smaller carrying load of AAV limits its use to packaging shorter transgenes. 29 The size of the transgene with promoters determines its packaging into viral vectors. Adenoassociated viruses are deemed the least toxic for in vivo use compared with AdV or LV, but they have the lowest carrying capacity and can only deliver small transgenes (e.g. hFIX). Larger transgenes (e.g. hFVIII or β-globin) can be readily packaged into AdV or LV vectors with larger loading capacities but limited in vivo utility, particularly for in utero use. 31,[38][39][40][41] However, emerging strategies, such as the use of smaller but functional transgenes, 42 the evolution of a dual AAV system, 43 or even an all-in-one AAV system 44 have provided solutions for the use of the less immunogenic AAV vectors.
Non-viral vectors do not present the immunotoxic or genotoxic adverse effects of viral vectors, but tend to be far less efficient at transducing target cells as they lack viral cell entry mechanisms; additionally, most techniques are suitable only for ex vivo cell transduction (e.g. lipofection, nucleofection, and plasmids) though the expanding library of lipid nanoparticles (LNP) designed for programmed release, slow release, and to display specific cell tropism, expands opportunities to use non-viral technology for in vivo strategies. 45 Gene therapy vectors are selected for each strategy based on carrying load (larger capacities are required for larger transgenes), tissue tropism (vectors can be designed to show greater affinity for certain cell types), immunogenicity, integration (needed for rapidly dividing stem cells which will lose episomal transgenes with cell division), cell toxicity, and insertional mutagenesis. Editing tools can be packaged in the same way as transgenes but require vectors with larger carrying capacities, such as AdV, LV, or LNP. Important factors by which to measure the success and safety of the chosen gene modification strategy are transduction efficiency (percentage of transduced cells in the target organ), transgenic protein expression, loss of protein expression with growth, immunemediated clearance of transduced cells, off-target and on-target integration, mutations and indels, germline mutation, transplacental trafficking of vector, and bystander effects in the mother. We will better understand the applicational benefits and challenges by looking at specific examples and preclinical models. A discussion of in utero stem cell transplantation is beyond the scope of this review.

| Haemophilias
The most common coagulation disorders are X-linked haemophilia A (Factor VIII deficiency affecting 1:5000 males) and haemophilia B (Factor IX deficiency affecting 1:35,000 males) and deficiencies of von Willebrand factor (affecting 1% of the population). The haemophilias are ideal conditions to treat in utero, considering the challenges both of severe disease and of conventional treatment. 46 Severe bleeding diathesis is associated with physiological deficiency of coagulation Factor VIII (FVIII) or IX (FIX), respectively, and is marked by spontaneous bleeding (<1% activity), which improves greatly following a modest increase to 3%-5% activity (mild-moderate disease and improved quality of life). Factors VIII and IX do not require specific organ expression; though the liver is the main protein factory, ectopic production has not proven detrimental to the treated subject. 46 Both factors have a wide therapeutic window and supraphysiological expression levels have not demonstrated toxicity in gene therapy recipients. 47,48 A major adverse effect of recombinant FVIII replacement is inhibitor production in~30% of patients. IUGT can induce tolerance to transgenic FVIII and may circumvent inhibitor formation. These factors lower the barrier for effective gene therapy solutions, which when applied in utero can potentially prevent perinatal intraventricular haemorrhage, umbilical cord haematoma, and other birth trauma associated with severe diseases. 49 Haemophilia A has initially proven a challenge to correct in gene therapy trials despite modest success in murine and canine models in which phenotypic correction has been achieved following systemic administration of AdV, LV, and AAV. [50][51][52][53][54][55] However, this appears to have been addressed with the development of a functional B-domaindeleted human-factor VIII that can be packaged into AAV5 or AAV3subtype 3 vector, demonstrating clinical efficacy. 42,56 Evidence of IUGT efficacy is limited: a sheep model replicating the spontaneous bleeding phenotype has been described, 57   We used a NHP model to demonstrate the clinical translatability of this approach, achieving supraphysiological hFIX expression in NHP recipients using AAV8 and AAV5, both of which are livertropic, 16 Importantly, we demonstrated the feasibility of a single intrauterine dose, given that transgene expression was maintained in most animals for several years with minimal neutralising response to transgene or AAV, and minimal immunotoxicity at the doses used.
Through multi-year surveillance of IUGT NHP, we demonstrated that transgene expression is related to serotype, GA, and gender of recipients. 17,72 Human FIX and FX were both expressed at therapeutic levels using the same AAV vectors, which were maintained over at least 50 months, despite rapid infant growth. 16

| Haemoglobinopathies
The haemoglobinopathies are the most common monogenic condi- Persistent correction of anemia suggested that fetal HSC were effectively targeted, and phenotype correction was achieved with <10% editing frequency.

| NEURODEGENERATIVE DISEASE
Neurogenetic disorders may manifest prenatally causing extensive and early-onset brain injury; examples include the mucopolysaccharidoses (MPS) and Gaucher type-II (acute infantile neuronopathic disease). Postnatal enzyme replacement efficacy is limited by the increasingly impermeable blood-brain barrier and irrecoverable neuronal damage. 96 An effective and safe gene therapy that can correct the underlying mutation, prevent early-onset neuropathy, and preserve the developing brain is urgently needed. 97  -679 glucocerebrosidase expression. 19 While untreated knockout mice exhibited fatal neurodegeneration and severely truncated lifespan, fetal IUGT recipients survived until at least 18 postnatal weeks, showed normal non-spasmodic mobility, and brain distribution of microglia, astrocytes, and neuronal lysosomes was restored. While the use of IUGT in enzyme deficiency disorders has yet to enter clinical trials, an interesting similar approach uses in-utero enzyme replacement therapy to treat Pompe's Disease, another enzyme deficiency disease, which has led to improved phenotype in utero and better outcomes at 13 months of life. 105 Another example of a life-threatening neurodegenerative disease amenable to IUGT is SMA, caused by mutation or deletion of the survival motor neuron 1 gene (SMN1) on chromosome 5q13, resulting in the depletion of the ubiquitous SMN protein required for multiple fundamental cellular homeostatic and bioenergetic pathways. 106 This affects 8-10/100,000 newborns globally per annum, and enhanced neuronal death is already detectable in utero. 107,108 Intracerebroventricular AAV9 delivering the human SMN to peripheral motor neurons in a murine knockout model of SMA extensively transduced various neuronal areas of the CNS with a great quantity of these being neural stem cells. 109 Clinical correction of muscular atrophy and neurological symptoms, with increased lifespan and improved postnatal growth, was observed in fetal IUGT mice. Intraperitoneal injection resulted in more widespread and less CNS-concentrated cellular transduction, and SMN transgenic protein expression was higher in the brain and spinal cord than in wild-type mice. This therapy has now been developed by a biotech company into an AAV vector sold commercially as Zolgensma®, but at a cost of €2 million even for a small child, the medication is out of reach for many families in need. 110 These experimental models give us a powerful demonstration of the potential benefits achievable with a single IUGT dose, particularly in a perinatally lethal or devastating condition without effective postnatal replacement therapy.

| CONGENITAL PULMONARY DISEASE
IUGT has successfully targeted monogenic pulmonary diseases, including congenital surfactant deficiency and cystic fibrosis. Intraamniotic delivery of GMT tools is ideal to target the developing lungs, primarily the pulmonary epithelial cells. 69,111,112 This will be a useful strategy for cystic fibrosis and inherited surfactant protein syndromes caused by mutations in surfactant system genes (SFTPB, SFTPC, or ABCA3), resulting in respiratory failure at birth, perinatal death, or chronic diffuse lung disease. 113 Cystic fibrosis is a multisystem disease caused by exonic and intronic mutations in the cystic fibrosis transmembrane regulator CFTR gene, resulting in the dysregulation of cAMP-regulated chloride channels controlling salt and fluid homeostasis in epithelial cells; this produces viscid secretions and epithelial cell dysfunction and ultimately mortality from respiratory failure. 114 Alleviating respiratory disease may substantially improve quality of life and decreased mortality.

| METABOLIC DISEASES
The liver is the producer of multiple proteins. Often, loss-offunction mutations cause aberrant protein formation or protein deficiency, resulting in abnormal metabolic processes, which may even manifest in utero, such as the lysosomal storage diseases. 118 Occasionally, loss-of-function mutations can benefit the individual, particularly for a disease caused by pathological accumulation of metabolites, such as in hereditary tyrosinaemia (HT1), a severe childhood disease characterised by accumulation of tyrosine and its byproducts in the liver, kidneys, and CNS causing organ failure and mental retardation. 119 Conventional treatment includes avoidance of tyrosine-containing foods and nitisinone, a hydroxyphenylpyruvate dioxygenase (HDP) inhibitor. 120   It is generally agreed that the high expenses and associated risks of GMT can be mitigated by a three-pronged approach: treating patients at a younger age and in better physical health, increasing the number of patients receiving therapy, and using the in vivo approach preferentially, 110,136 all of which are represented in an intentional IUGT strategy. There should be an international consensus on the highest priority diseases to target, which may vary geographically, as gene therapy products are expensive to produce and require lengthy validation of efficacy and safety according to good manufacturing practice regulations. 13,137 Numerous factors influence the decision to proceed to clinical applications, such as population carrier frequency, affordability of conventional medical care as well as the priority placed on advanced therapeutics, and ethical considerations will associate closely with societal acceptance of genetic diseases, pregnancy termination, and pregnancy interventions. Focusing research and clinical efforts on the more common diseases like the haemoglobinopathies, SMA, and cystic fibrosis may increase the likelihood of reaching global clinical trials of IUGT in the near future.
Unresolved issues need to be addressed through improving vector design and longitudinal surveillance in large animal models, and these include long-term disruptive effects of vector integration, genotoxicity and malignant change, future reproductive health, and long-term maternal off-target effects. The time to step into clinical trials is fast approaching, and fetal medicine providers and gene therapists should function within consensus-lead guidelines to carefully, but boldly, step into this new therapy.