The Future of In Utero Gene Therapy

Abstract

Significant advances in the safety and efficacy of gene therapy have sparked a new frontier in therapeutics for genetic diseases as evidenced by the greater than 700 active gene therapy investigational new drug applications reported by the NIH and the US Food and Drug Association [1]. Although postnatal gene therapy trials are encouraging, limitations to effective therapy including an immune barrier and initiation of treatment after disease onset can exist. Advances in prenatal diagnostics provide hope that many genetic abnormalities will be able to be diagnosed before birth. Prenatal gene therapy has the potential to take advantage of normal developmental properties of the fetus and overcome some of the current limitations to efficient postnatal gene therapy. In the current review, we discuss the rationale for prenatal gene therapy including the small fetal size, the tolerogenic fetal immune system, the presence of highly proliferative and accessible stem/progenitor cells of multiple organs and, ultimately, the ability to treat diseases in which irreversible pathology begins prior to birth. This rationale is discussed in the context of published animal studies. We highlight the ethical considerations that are unique to prenatal gene therapy, namely the importance of rigorous evaluation of fetal germ cells and developing organs as well as the mother noting that animal studies to date have not demonstrated any significant germline or maternal effect of prenatal gene therapy. Finally, the practical considerations of future clinical prenatal gene therapy are discussed including potential initial target disease characteristics and the importance of non-directive prenatal counseling of families carrying a fetus with a genetic diagnosis.

1. Introduction

Single gene mutations (monogenic) are responsible for a large number human diseases, many of which result in significant morbidity and mortality (World Health Organization). Advances in genetic diagnostic capabilities allow for the identification of the affected gene and causative mutations of many of these diseases. As such, gene therapy including gene editing approaches, to treat these diseases at their genetic “roots” is an attractive therapeutic option. Broadly speaking, traditional gene therapy involves gene augmentation via viral or nonviral mediated delivery of a therapeutic gene. Gene editing approaches involve the use of nucleases including zinc-finger nucleases, TALENs (transcription activator-like effector nucleases) and CRISPR-Cas9 (clustered regularly interspaced short palindromic repeats [CRISPR]-CRISPR-associated 9) to instigate a sequence-specific change in the DNA. Furthermore, gene therapy can also be broadly classified as ex vivo – treatment of cells ex vivo with the subsequent transplantation of corrected cells – or in vivo – delivery of the gene therapy directly into the patient. To date, there have been no clinical trials of in utero gene therapy and most research has focused on viral vector mediated therapeutic gene augmentation following in vivo delivery. More recently, early proof-of-concept studies have been performed evaluating in utero somatic cell gene editing in mouse models of human diseases via viral and nonviral delivery mechanisms. As with all gene therapy approaches, both prenatal and postnatal, the safety and efficacy of these treatments must be rigorously established in animal models prior to clinical translation. This “safety bar” becomes even more paramount for an in utero approach in which the therapy is provided to a fetus during crucial points of development and in which, by nature, there are two patients – the mother and the fetus. In this review, we will discuss the rationale for in utero gene therapy, review some of the preclinical animal studies in the context of this rationale, describe some of the recent studies involving in utero gene editing and an ex vivo approach to prenatal gene therapy, and highlight some of the ethical issues in the context of characteristics of potential initial target diseases for in utero gene therapy.

2. In utero gene therapy: Rationale and supporting animal studies

The developing fetus has many innate properties that potentially make it an ideal recipient for in vivo gene augmentation and/or gene editing (Figure 1). The small size of the fetus allows one to maximize the gene therapy dose per recipient weight. Studies of in utero hematopoietic cell transplantation (IUHCT) have demonstrated the ability to safely deliver hematopoietic progenitor cells intravascularly into a 16 week gestation fetus. Similarly, intravenous blood transfusions via umbilical vein injection for Rh disease are routinely performed at 18 weeks’ gestation with minimal procedural risk [2]. Similar techniques could be used for intravascular delivery of gene therapy at these gestational ages. At this age, the average fetal weight is ~150 grams in contrast to a 4 kilogram newborn or a 60 kilogram adult (1:27:400 weight ratio) providing a significant dose advantage as well as minimizing constraints on large-batch viral vector production for a fetal recipient.

Figure 1.

The rationale for in utero gene therapy.

In addition to the small fetal size, the immune system of the developing fetus is more prone to a tolerogenic phenotype [3]. A current limitation to postnatal gene therapy is the presence of a preexisting immune response to the viral vector or transgene product and/or the elicitation of a neutralizing immune response to either the viral vector or transgene product following gene therapy [4-9]. Specifically, anti-AAV antibodies secondary to natural exposure have been documented in individuals as early as the newborn period and the presence of these antibodies would preclude a patient from a gene therapy trial using that AVV serotype [5]. Presumably, pre-existing maternal anti-AAV antibodies could cross the placenta to limit in utero gene transfer [10]. This limitation could be avoided by screening the mother for anti-AAV capsid antibodies or by performing the therapy in the first half of the second trimester as the transport of maternal IgG into the fetal blood begins in the second half of the second trimester and peaks during the third trimester [11]. Similarly, preexisting antigen specific T cells and antibodies to Streptococcus pyogenes Cas9 (SpCas9) and Staphylococcus aureus Cas9 (SaCas9) have been identified in the peripheral circulation of individuals presumably from a previous bacterial infection [12]. This adaptive immune response has the potential to limit the efficacy of postnatal in vivo CRISPR-Cas9 gene editing therapy. Finally, even if the absence of a preexisting adaptive immune response, gene therapy can elicit an immune response to both the viral vector and transgene limiting the efficacy of future repetitive doses which may be necessary [9]. In contrast, multiple animal studies have demonstrated the induction of immune tolerance to the transgene product following prenatal gene therapy which has allowed for repetitive doses after birth [13-15]. This is exemplified in the studies of in utero gene therapy in the mouse model of hemophilia B. In this study, persistent expression of human factor IX (hFIX) in the absence of any cellular or humoral immune response to hFIX was achieved following in utero intramuscular injection of AAV1 carrying the hFIX transgene (AAV1.hFIX) with subsequent postnatal readministration of AAV1.hFIX. In contrast, adult recipients of AAV1.hFIX developed antibodies to hFIX [15]. More recently, AAV5 or AAV8 vectors encoding hFIX or hFX was delivered in utero in a nonhuman primate model at a gestational age equivalent to 12 to 14 weeks gestation in humans [13]. This study demonstrated persistent curative levels of hFIX and hFX in the circulation, durable immune tolerance to the hFIX, and no evidence of clinical toxicity. However, in contrast to transgene tolerance, animal studies suggest that tolerance to the viral vector is not achieved following in utero delivery, likely related to the transient presence of the viral vector capsid proteins in contrast to the transgene which maintains persistent expression [14, 15]. Thus, if multiple repeat administrations of the gene therapy is believed to be necessary following an initially in utero approach, use of alternative vector serotypes and/or mechanisms to induced tolerance to the viral vector need to be explored.

Paramount to successful gene therapy is the ability to efficiently deliver the transgene to the target organ and target cell population within that organ. Often, the target cell population is a stem cell or progenitor cell. This is particularly true for integrating viral vectors and/or gene editing in which DNA modifications of a stem/progenitor cell will be passed on to subsequent daughter cells with the hope of providing therapeutic genetic correction for the lifetime of the individual. In contrast to most postnatal organs, stem/progenitor cells of multiple organs in the developing fetus are abundant and readily accessible. The accessibility of different organs/cells to viral vector transduction is dependent on multiple factors including the vector serotype as well as the route of vector delivery and the developmental stage at the time of injection [16-21]. This is exemplified by mouse studies targeting the lung and skin. Intraamniotic injection of viral vectors late in gestation, after fetal breathing movements have begun, provides a relative specific means of targeting the pulmonary epithelium via “inhalation” of the amniotic fluid [17, 21, 22]. Alternatively, intraamniotic injection of viral vectors early in gestation prior to the formation of a protective keratinized skin layer has been demonstrated to efficiently target skin progenitor cells [16, 23]. Additional routes of injection include intramuscular, intraventicular within the brain to target the CNS and intravascular via the vitelline vein in the mouse model which drains directly into the portal circulation. In large animal models and humans alternative, minimally invasive systemic delivery approaches would include ultrasound-guided intraperitoneal, intracardiac and umbilical vein injection as well as ultrasound-guided intraventicular injection to target the brain and ultrasound guided or minimally invasive intratracheal or intra-lung parenchymal injection to target the lung. Furthermore, genetic diseases of the central nervous system may benefit from the more permissive blood-brain barrier present in the fetus such that efficient brain-targeting could be achieved following systemic viral vector delivery as has been demonstrated in animal models with AAV9 [19, 20]. Finally, in addition to being accessible, fetal stem/progenitor cells of multiple organs are often more proliferative in contrast to postnatal stem/progenitor cells which tend to exist in a quiescent state under homeostatic conditions. As discussed below, this may provide a significant advantage to increase the efficiency of CRISPR-Cas9 mediated homology directed repair (HDR).

One of the strongest rationales for in utero gene therapy is to instigate a curative treatment before birth. This is an attractive option for diseases that are lethal in the prenatal or perinatal period or in which irreversible disease pathology begins prior to or immediately after birth. This benefit of in utero gene therapy is exemplified by lysosomal storage diseases which may be fatal before birth such as mucopolysaccharidosis type VII or in which irreversible disease pathology begins in the fetus although the patient may appear phenotypically normal at birth [24]. Neuronopathic Gaucher disease is a lysosomal storage disease caused by a deficiency of the glucocerebrosidase enzyme. Irreversible neurologic pathology begins in utero and the acute childhood form of the disease is lethal. Since replacement enzymes cannot cross the blood-brain barrier, neuronopathic Gaucher disease has no effective therapy. In a recent study, in utero delivery of the glucocerebrosidase gene via intracranial injection of AAV9 resulted in neuronal glucocerebrosidase expression, mitigation of neuroinflammation and neurodegeneration and survival of mice with normal mobility up to 18 weeks of life [25]. Finally, due to the changing physiologic function of some organs at the time of birth, diseases which do not cause a problem before birth may become lethal at the time of birth. This is best exemplified by Surfactant protein B deficiency, a rare autosomal recessive congenital lung disorder which is rapidly fatal from respiratory failure in the newborn period and for which the only treatment is a lung transplant [26]. Genetic correction of this disease before birth, when lung function is not required, has the potential to provide adequate time for normal lung development before birth and adequate lung function at the time of birth.

3. Alternative approaches: In utero somatic cell gene editing and ex vivo in utero gene therapy

In utero somatic cell gene editing.

Although the majority of studies of in utero gene therapy have focused on “traditional” gene augmentation, recent advances in gene editing technology, including CRISPR-Cas9, have galvanized excitement for the potential of gene editing before birth to provide a permanent curative change in a gene after a single treatment for some genetic diseases. A detailed review of all gene editing approaches is beyond the scope of this review however we will briefly highlight some the basic aspects of CRISPR-mediated gene editing. Standard CRISPR-Cas9 editing generates a double strand DNA break (DSB) that is usually repaired by non-homologous end-joining (NHEJ). Insertion and deletion of bases (indels) at the site of NHEJ can silence a gene or, if two guide RNAs (gRNAs) are used to target the gene at two sites, a gene or segment of DNA can be excised. Alternatively, if a DNA repair template is provided with Cas9 and the gRNA, homology-directed repair (HDR) can correct the target mutation via incorporation of the repair template at the time of repair. NHEJ is the more efficient pathway and, as alluded to above, HDR requires proliferating cells to occur efficiently [27]. Finally, base editors consist of a catalytically impaired CRISPR-Cas9 fused to either a cytosine or adenine deaminase domain [28-31]. Base editors can make site-specific base changes (C-to-T for cytosine deaminase; A-to-G for adenine deaminase) without the requirement for double-strand DNA breaks or proliferating cells. Studies of in vivo postnatal gene editing are encouraging but highlight potential limitations to the postnatal approach including an immune barrier, low HDR editing efficiency due, in part, to the lack of accessible proliferating target cells, and the inability to treat diseases that are morbid or cause death in the perinatal period [32-36]. In utero gene editing has the potential to overcome these limitations based on the rationale outlined above.

To date, a limited number of in utero somatic cell gene editing studies have been published and have established the initial feasibility of this approach. In two of these studies adenoviral vectors were used to deliver SpCas9 or a SpCas9-based base editor to fetal recipients in mouse models of a human metabolic liver disease and a congenital lung disease. Specifically, CRISPR-NHEJ was shown to efficiently edit pulmonary epithelial cells before birth and improve survival in a lethal-at-birth mouse model of the congenital lung disease, surfactant protein C deficiency (Figure 2) [22]. In another study, base editing with the widely used base editor 3 (BE3) efficiently introduced a nonsense mutation in the Hpd gene in hepatocytes of the tyrosinemia mouse model. Silencing of the Hpd gene rescued the lethal phenotype of this model by preventing the accumulation of toxic metabolites in the tyrosine catabolic pathway (Figure 2) [37]. Hepatocytes edited prenatally were noted to persist up to 3 months post editing (the last study time point) in contrast to those edited postnatally which decreased after 1 month. The loss of edited cells was associated with elevated circulating levels of anti-adenoviral and SpCas9 antibodies compared to prenatally edited mice likely related to the tolerogenic environment of the fetus. Importantly, in these initial studies, no editing was noted in the germline of fetal recipients or in maternal organs, two areas that must be rigorously investigated in small and large animal models prior to any thought of clinical application. Furthermore, these initial studies used an adenovirus to deliver the editing constructs secondary to its large packaging capacity and its ability to efficiently transduce target cells. Additional studies investigating more clinically relevant delivery mechanisms are of utmost importance as one considers any possible clinical application of this approach in the future. A third study used nanoparticle-mediated delivery of peptide nucleic acids (PNAs) and a donor template to target hematopoietic progenitor cells (HPCs) before birth for correction of the pathologic mutation in the β-globin gene in a mouse model of human β-thalassemia [38]. This approach uses HDR and noted an improvement in disease phenotype associated with a 6% and 10% mutation correction rate in total bone marrow cells and isolated HPCs respectively after a single in utero treatment. This stood in contrast to an editing efficiency of 4% and 7% in total bone marrow cells and HPCs respectively after 4 postnatal treatments highlighting the potential benefit for HDR of increased proliferating target cells before birth for HDR [39].

Figure 2. Feasibility of in utero gene editing targeting the fetal lung and liver.

Recent studies have demonstrated the feasibility of prenatal CRISPR-mediated gene editing targeting the lung [20] (A,B) and liver [35] (C,D). (A,B) Adenoviral vectors carrying SpCas9 and a loxP-targeting guide RNA (gRNA) were injected into the amniotic cavity of gestational day (E) 16 mTmG mice. Successful editing is indicated by expression of green fluorescence. Flow cytometry demonstrated that pulmonary epithelial cells were the predominant cell type edited (A) and immunohistochemistry (IHC) suggested that alveolar type 1 (AT1), alveolar type 2 (AT2), secretory, and ciliated epithelial cells were edited (B). (C,D) Adenoviral vectors carrying a cytosine base editor and Hpd-targeting gRNA were intravascularly injected into E16 hereditary tyrosinemia type 1 (HT1) fetuses. Successful editing introduced a nonsense mutation in the Hpd gene resulting in decreased protein expression on IHC (C) and survival similar to HT1 mice maintained on the drug NTBC which inhibits the HPD enzyme (D).

Finally, it is important to draw the distinction between mid-to-late gestation somatic cell in utero gene editing and preimplantation embryonic gene editing. While the mutation must be known prior to implantation for preimplantation embryonic gene editing, in utero gene editing could be applicable to de novo mutations or mutations diagnosed later in pregnancy. Early studies suggest that in utero gene editing does not affect the germline in contrast to preimplantation embryonic gene editing in which the goal is often to correct a mutation in the germline. In utero gene editing, similar to augmentation gene therapy, can be directed to specific organs and/or cell types by various modes of delivery and viral vectors while preimplantation embryonic gene editing has the potential to affect all cells of the body. Finally, preimplantation embryonic gene editing is performed ex vivo and thus has little risk of affecting the mother in contrast to in utero gene editing which is an in vivo approach and thus maternal implications must be considered.

Ex vivo in utero gene therapy.

Ex vivo gene therapy in which autologous cells are harvested from a patient, targeted for augmentation gene therapy/editing in vitro, and then transplanted back into the patient following a conditioning regimen is a promising approach for genetic diseases, including hematologic diseases, in the postnatal setting [40]. This approach typically employs correction of autologous CD34⁺ HPCs and has the benefit of not exposing nontarget cells to the viral vector and, in the case of prenatal gene therapy, not exposing the mother to the therapy. In the fetus, an autologous source of HPCs could theoretically come from the fetal liver/peripheral blood, the amniotic fluid [41] or the placenta. However, application of this approach to the fetus poses two main considerations: 1. it is not possible to safely harvest an adequate number of autologous fetal liver- or fetal peripheral blood/umbilical vein-derived CD34⁺ cells or expand and maintain hematopoietic differentiation of amniotic fluid- or placental- derived stem cells for subsequent correction and transplantation; and 2. use of a third party source of HPCs is possible but these cells would presumably have the corrected gene and thus no additional ex vivo gene therapy would be required. This second consideration, an in utero third party hematopoietic cell transplantation, holds clinical promise for a number of congenital hematologic diseases such as Sickle cell disease and thalassemia and is currently under investigation [42]. An alternative approach of ex vivo gene therapy, however, has been identified that is applicable in utero and has the potential to provide a cure for diseases resulting from a deficit in a secreted protein. Specifically, Almeida-Porada et al. have transplanted human c-Kit⁺ placental cells transduced ex vivo with a lentivirus containing the FVIII transgene into the peritoneal cavity of fetal sheep as a therapeutic strategy for hemophilia A. They demonstrated a 30-237% increase in plasma FVIII activity at 4 months after transplant (1 month of age) that was stable at 5 months post-transplant despite a 3-fold increase in plasma volume [43, 44]. These studies are encouraging as a potential off-the-shelf therapy for hemophilia A. However, successful clinical translation of this approach for hemophilia A and other diseases will be dependent on achieving persistent engraftment of an adequate number of cells to produce high enough levels of the desired protein.

4. Ethical and practical considerations

In utero gene therapy has the potential to cure or mitigate a large number of genetic disorders that currently have ineffective treatments. However, there exist ethical and practical considerations for a prenatal gene therapy approach that are not present for postnatal gene therapy. As with any fetal intervention, two patients are involved, the mother and the fetus, and the timing of the intervention during development raises important considerations. A recent panel discussion with international experts was held at the International Fetal Transplantation and Immunology Society (IFeTIS) annual meeting and summarized many of these pertinent considerations [45]. Specifically, with respect to the safety of in utero gene therapy, it is important to note that complications such as infection, preterm labor, and fetal loss are all theoretically possible, however, strong clinical data exists supporting the ability of a fetus to tolerate multiple minimally invasive accesses via the intravascular, intraperitoneal, or intraamniotic routes with very low procedure-related risks. As alluded to above, maternal safety is of utmost importance. Continued animal studies investigating the possibility of maternal exposure to the viral vectors and transgene products including any potential maternal immune response to the injectate need to be performed. Finally, for prenatal in vivo gene augmentation and gene editing, the frequency and effect of integration of the viral vector and/or off-target effects must be clearly defined prior to clinical application. Encouragingly, data in animal studies thus far have not demonstrated significant germline integration/editing or any unintended off-target effects [22, 37, 38]. These off-target analyses for in utero gene editing have been performed in a biased fashion and thus, future, rigorous, unbiased analyses are warranted.

5. Conclusions

Over the past 20 years significant advances in postnatal gene therapy have been made with clinical trials now providing hope for patients with once untreatable genetic diseases. Over this time, our understanding of the potential benefits and risks of prenatal gene therapy has expanded through studies in animal models. With an eye towards clinical application in the future, additional practical and ethical considerations involve determining the characteristics of the ideal initial target diseases and clinical application of prenatal gene therapy. Ideal diseases include those that result in severe morbidity and/or mortality either in utero or after birth and in which irreversible pathology begins before or shortly after birth and for which effective postnatal therapies do not exist. Candidate diseases should have a reliable prenatal enzymatic or genetic diagnosis and strong genotype/phenotype correlation. Ideally, an animal model of the disease should exist in which in utero gene therapy has been demonstrated to be safe and efficacious. Finally, it is paramount to recognize and respect the different beliefs and values families hold with respect to pregnancy and the diagnosis of a fetus with a congenital disorder. As such, future clinical application of in utero gene therapy will mandate non-directive counseling of families in which the options of no prenatal therapy and experimental prenatal gene therapy, including all of the possible risks and benefits, are provided without bias and the decision of the family is supported. Once prenatal gene therapy is demonstrated safe for the treatment of the initial target diseases and the approach to counseling families and implementation of in utero gene therapy is mastered, expansion of in utero gene therapy more broadly could be considered.

KEY POINTS.

1. In utero gene therapy, including novel gene editing approaches, has the potential to treat genetic diseases before the onset of irreversible pathology and rescue diseases that are perinatal lethal or which cause significant morbidity and for which no effective treatment exists.

2. In utero gene therapy takes advantage of normal developmental properties of the fetus including small fetal size, a tolerogenic immune system and accessible/highly proliferative stem/progenitor cells of multiple organs to achieve efficient persistent target gene correction/augmentation.

3. Unique safety and ethical considerations exist for in utero gene therapy related to the prenatal timing of the therapy and the fact that two patients are involved – the mother and the fetus. Animal studies to date have not demonstrated a significant effect of prenatal gene therapy on fetal germ cells or maternal organs.