Mechanistic Insights into Factor VIII Immune Tolerance Induction via Prenatal Cell Therapy in Hemophilia A

Abstract

Purpose of Review:

Prenatal stem cell and gene therapy approaches are amongst the few therapies that can promise the birth of a healthy infant with specific known genetic diseases. This review describes fetal immune cell signaling and its potential influence on donor cell engraftment, and summarizes mechanisms of central T cell tolerance to peripherally-acquired antigen in the context of prenatal therapies for Hemophilia A.

Recent Findings:

During early gestation, different subsets of antigen presenting cells take up peripherally-acquired, non-inherited antigens and induce the deletion of antigen-reactive T-cell precursors in the thymus, demonstrating the potential for using prenatal cell and gene therapies to induce central tolerance to FVIII in the context of prenatal diagnosis/therapy of Hemophilia A.

Summary:

Prenatal cell and gene therapies are promising approaches to treat several genetic disorders including Hemophilia A and B. Understanding the mechanisms of how FVIII-specific tolerance is achieved during ontogeny could help develop novel therapies for HA and better approaches to overcome FVIII inhibitors.

Introduction

Factor VIII Function and Genetic Pathogenesis of Hemophilia A.

Hemophilia A (HA) is a monogenic, X-linked disorder caused by a deficiency in coagulation factor VIII (FVIII), a vital procofactor in hemostasis. Mature FVIII is a heterodimeric (Heavy chain: A1-a1-A2-a2-B, Light chain: a3-A3-C1-C2) glycoprotein comprised of 2332 amino acids (1, 2). Normal levels of FVIII coagulant activity (FVIII:C) in plasma range from 50–150 IU/dL, but since FVIII circulates in the bloodstream bound to von Willebrand Factor (vWF), FVIII’s concentration is influenced by vWF levels, and factors affecting them, such as ABO(H) blood group and hemodynamic shear forces (3–6). Patients with severe HA have FVIII:C levels of less than 1 IU/dL, and as a result experience spontaneous bleeding ranging from subcutaneous hematomas to life-threatening intracranial hemorrhaging (7, 8). Patients with moderate HA (1–5% FVIII:C) experience most bleeding following injuries, but also have spontaneous bleeding events into their joints and muscles, while mild HA patients with more than 5% FVIII:C only sustain bleeding events after serious trauma or surgery.

Mutations associated with severe HA are diverse and can cause a spectrum of dysfunctional FVIII proteins or its complete absence, and they include inversions in intron 22 (~45%), frameshift deletions/insertions (~16%), missense mutations (~15%), premature termination mutations (~10%), single/multi-exonic deletions (~3%), splice site mutations (3%), inversions in intron 1 (2%), and other, as-yet, undefined mutations (9–11). By contrast, mutations in non-severe HA are usually missense mutations (12). Because FVIII inhibitors represent antibody-mediated responses to the missing protein, the risk of their development is related to and can be predicted from the type of mutation (13). In addition to the type of F8 mutation (9), IL-10 and TNFα-encoding gene polymorphisms (14–16), ABO(H) blood group (3, 17), African ancestry (specific to Int22 inversion (18)), high-dose (≥150 IU/kg/week) FVIII infusion during early treatment or surgery (19–22), and type of factor concentrate administered during early treatment (23–25) have all been associated with inhibitor formation.

Since HA is an X-linked disease, it typically affects males. Nevertheless, the disease can also manifest in females carrying mutations in one or both alleles of the F8 gene; severe HA can occur in female carriers due to X-inactivation of the normal F8 allele, or through inheritance of an aberrant F8 allele from both parents (26–29).

Current Standard of Care for HA and Novel Therapeutic Approaches.

Currently, the standard of care for severe HA in western countries consists of prophylactic FVIII replacement therapy using plasma-derived FVIII (pdFVIII) products and/or recombinant FVIII products (rFVIII), with standard or extended half-life (30–34). While FVIII therapy has allowed HA patients to have near normal life expectancies, the development of anti-FVIII neutralizing antibodies (FVIII inhibitors) in up to 30% of patients continues to be a major complication, rendering replacement therapy ineffective and increasing risk of morbidity and mortality (35, 36). Furthermore, the sustainability of the current Standard of Care model is far from ideal, as the HA treatment costs are estimated at an average of $300,000 per year/patient (median), while for patients with inhibitors the cost triples (37, 38).

Even though administration of very high doses of FVIII can be used in certain individuals, HA patients who develop antibodies are extremely difficult to manage. Therefore, products such as RNA interference therapeutics (39), FVIIa variants (40), and FVIII-bypassing agents have been developed that are able to circumvent the need for the deficient factor and thereby restore hemostasis.

Among FVIII-bypassing agents, a bi-specific antibody to FIX/FIXa and FX/FXa has become available for use in HA patients without inhibitors (41). Advantages of this bi-specific antibody include the estimated half-life of 2 to 4 weeks, subcutaneous administration, and the minimal risk of inhibitor formation (41–43). However, a recent study demonstrates that FIX/FIXa/FX/FXa bi-specific antibody lacks binding specificity between zymogen and enzyme (i.e. FIX vs FIXa), which causes the reaction-limiting molecule to be FIXa, instead of FVIIIa, and does not promote platelet surface binding (44). In addition patients that received FIX/FIXa/FX/FXa bi-specific antibody in combination with activated prothrombin complex concentration or FVIIa, are at increased risk for thrombotic complications, an atypical outcome in standard HA therapy (45).

In the recent years, gene therapy (GT) has emerged as a promising approach for severe HA. In the US, there are currently 6 ongoing GT clinical trials for the management of HA, all of which are using AAV-based vectors to deliver the F8 transgene (46). In the first of these, a Phase I/II dose escalation study (BioMarin; BMN 270) [ClinicalTrials.gov Identifier: NCT02576795] involving 9 severe HA patients, results to date have shown that a single administration of an AAV serotype 5 (AAV5) vector encoding a hF8 transgene under the control of a liver-specific promoter can yield durable circulating FVIII:C levels within the normal/healthy range (>50 IU/dL) (47). Specifically, one-year follow-up has shown normal FVIII:C levels in patients who received a single high-dose (6×10¹³ vector genome per kg [vg/kg]) of BMN 270, while patients in the low-dose and intermediate-dose cohorts had circulating FVIII:C levels that peaked at only 3 IU/dL. While no AAV5 capsid-specific immune responses nor FVIII inhibitors were reported, elevated transaminase levels in the first high-dose patient led physicians to implement prophylactic prednisolone in the remaining patients. Despite this auspicious start, however, available three-year data suggest that the plasma FVIII levels are declining and the patients may have to be treated again in the future to maintain therapeutic levels.

In another Phase I/II study (48) (SPK-8011; Spark Therapeutics) [ClinicalTrials.gov Identifier: NCT03432520] a novel liver-tropic AAV vector encoding a B-domain deleted (BDD) human F8 transgene was used. The study has shown up to 14% FVIII:C in its intermediate dose cohort of 5×10¹² vg/kg, thereby surpassing the conservative study goal of 12% FVIII:C that was assumed to be necessary to prevent spontaneous bleeding. However, 2 of the patients developed significant immune response to the AAV capsid, one of which required hospitalization. This anti-capsid response, and the steroid therapy administered to counter it, led to a drop in FVIII levels to below 5% in these patients (48).

Bayer also recently initiated a Phase 1/2 open-label safety and dose-finding study of their AAV therapeutic, BAY2599023 (DTX201), in 18 adults with severe HA [ClinicalTrials.gov Identifier: NCT03588299]. This trial began enrolling in November 2018, and no results have yet been reported. Shire has also launched a global Phase I/II dose escalation study (clinicaltrials.gov/ct2/show/NCT03370172) to use an AAV8 vector encoding a BDD F8 transgene (SHP654/BAX 888) in a total of 10 participants.

A cost-effectiveness analysis was performed to compare the treatment of severe HA via GT vs. by FVIII prophylaxis, and this analysis supported significantly lower per-person costs for GT over 10-year timeframes (49). Unfortunately, however, the presence of pre-existing immunity to AAV in as many as 60% of individuals (depending on donor age and AAV serotype) severely limits the number of patients who can potentially benefit from this type of therapy (50–52). Furthermore, after AAV vector exposure, patients may become immunized against the capsid and develop high titers of neutralizing antibodies (Nabs) that persist long-term, precluding patients from receiving another dose of an identical product (53). Administering the AAV vector during the prenatal or neonatal period could solve this problem, since treatment during this window of time would ensure the vast majority of patients are devoid of pre-existing immunity. However, little is currently known about the possible inflammatory responses and potential hepatoxicity that may result from vector-host interactions after in vivo administration of AAV vectors during these early developmental stages, highlighting the need for further carefully designed safety studies prior to implementing AAV-based GT in fetal/neonatal recipients (54). If, however, instead of directly injecting viral-based vectors into the recipient, suitable cells were modified in vitro and then used as vehicles to deliver the therapeutic transgene, a much greater degree of manufacturing control would be afforded, and GT could be applied to neonatal and even fetal recipients with far lower risk and improved safety. In the following sections, we provide a concise background on prenatal cell transplantation and then focus on the use of such an approach as a treatment for HA.

Prenatal Transplantation (PNTx) as a Treatment for Hemophilia A

PNTx holds great promise for treating/curing many inherited genetic diseases early in gestation, (55, 56). To-date PNTx has been performed on a compassionate-use basis in roughly 50 patients in an effort to treat 14 different genetic disorders (57–59). Specifically, regarding the use of PNTx to treat HA, HA is the most common inheritable coagulation deficiency (60) and 75% of HA patients have a family history. Thus, prenatal diagnosis is feasible, available, and strongly encouraged in most Western and developing countries (61–71). Digital PCR now allows analysis of free fetal DNA in maternal plasma, making non-invasive in utero diagnosis of HA possible as early as 7 weeks of gestation (70). Furthermore, prenatal diagnosis and family history can predict whether patients will have severe, moderate, or mild HA and their likelihood of developing inhibitors to FVIII. Of note, performing PNTx to treat HA would circumvent the immune barriers present in adult patients, and could induce immunological tolerance to therapeutic antigens, such as FVIII (72–77), and thereby eliminate the risk of inhibitor formation (78–82) if/when subsequent postnatal treatment is needed.

When considering a prenatal treatment, safety is obviously of the utmost importance. It is therefore critical to note that during early fetal life, activation of FX occurs predominantly via tissue factor activity, making it largely independent of the FIXa/FVIIIa phospholipid complex (83). As a result, the fetus develops without hemorrhage, despite having little or no expression of FVIII and FIX (83–85). The unique hemostasis of the fetus should thus allow prenatal treatment to be performed safely for HA; indeed, one of the 50 human patients that received PNTx was transplanted with fetal liver cells in an effort to correct HA, and although the patient was not cured, immune tolerance to FVIII was induced, and the patient did not developed the high-titer inhibitors from which the patient’s siblings had suffered (86). Taken together, these data provide compelling evidence that HA is a highly promising candidate disease for treatment prior to birth (reviewed in greater detail in (87, 88)).

Human Fetal Immune Development and the Unique Opportunity to Induce Tolerance

It was a long-held assumption that the fetal immune system was immature and the fetus was, for all intents and purposes, immuno-naïve since it develops in a sterile environment without exposure to foreign antigens. In recent years, however, there has been an ever-growing appreciation that the developing immune system is, in fact, in contact with many immune stimulatory molecules (89) such as maternal haploidentical antigens (90–92), microbes, materials present in the amniotic fluid, and food antigens (93). As a result, it is now acknowledged that although the immune system may only reach full maturity several years after birth, the immune system begins to develop and gains a surprising degree of functionality fairly early during gestation.

Looking at clinical studies data to date, it is interesting to note that, although the fetal immune system is known to allow the engraftment of haploidentical or allogeneic donor cells (94, 95), clinical success with PNTx using hematopoietic stem cells (HSC) has thus far been limited to recipients with immunodeficiencies, conditions in which donor cells would be predicted to possess a selective proliferative and/or survival advantage over the cells of the host (58, 96–99). Because HSC-PNTx is performed without myeloablation or immunosuppression (to avoid possible risks/toxicities to the fetus and mother), immunologic barriers and the absence of stress-induced signaling have been considered as significant contributors to the limited donor HSC engraftment (100–102) thus far seen in patients with diseases that would not be expected to confer a selective advantage to the transplanted donor cells. It is worth noting that the majority of human patients have, in fact, engrafted following HSC-PNTx, irrespective of their underlying disease, and the levels of engraftment, while not high enough to mediate disease correction, were sufficient to induce immune tolerance to the donor (100–102). Importantly, the induced state of tolerance was donor-specific, since the recipient maintained normal immune-reactivity to unrelated third-party antigens (100–102). Despite the obvious clinical importance, the specific pathways present within the fetus that make it possible to achieve a state of immune tolerance as a result of prenatal exposure to exogenous antigens are not currently well understood, although existing evidence suggests that adaptive immune cells of the fetus are likely to be the primary contributors to the pro-tolerogenic signaling that exists toward haploidentical cells present in fetal tissues during gestation.

In the following sections, we will provide an overview on the prevailing thoughts regarding some of the processes thought to play a role in maintaining the delicate yet critical balance between immunity and tolerance during fetal life and will review mechanisms by which PNTx is thought to promote the induction of immune tolerance to exogenous proteins and/or donor cells that are introduced to the fetus during gestation and place these data in the context of using PNTx to treat HA.

Induction of central tolerance during fetal development

The fetal immune system differs from that of the adult in that it lacks a robust response to foreign antigens, with fetal T cells being more likely to engage in promoting tolerance. During fetal life, T cell progenitors are educated in the thymus, a primary lymphoid organ, to ensure development of tolerance to self-antigen and non-inherited maternal antigen (NIMA) (103). Studies of human thymus organogenesis have shown that the first T-lineage affiliated CD34^int CD45⁺CD7⁺ hematopoietic progenitor cells enter the developing thymus by 8 gestational weeks (gw) (104), and can become CD4/CD8 double-positive (DP) thymocytes soon after (105). Concurrently, cortical thymic epithelial cells (TEC) and medullary TEC undergo sufficient differentiation to orchestrate a diverse array of thymocyte selection processes (104). Cortical TEC are the primary mediators of positive selection, a process during which DP thymocytes with low affinity/avidity for peptide-MHC are rescued from programmed cell death, undergo CD4 or CD8 single-positive T cell commitment, and begin to express elevated levels of CD69, an early activation marker (106). After chemokine-based recruitment to the medulla, the maturing thymocytes with low affinity become classified as naïve CD4⁺ or CD8⁺ T cells, and emigrate to the fetal periphery by 14–16 gw (107), while those exhibiting high reactivity to peripherally- or thymus-acquired antigen (Ag), presented on antigen-presenting cells (APC), undergo depletion. Alternatively, in the presence of medullary thymic epithelial-produced thymic stromal lymphopoietin (TSLP), dendritic cells (DC) resident within the medulla can instruct these high-affinity thymocytes to differentiate into natural T_reg cells (nT_reg) (108–111), whose repertoire is capable of establishing immunotolerance. Other APC subsets, including B lymphocytes, are also known to contribute to thymic central tolerance (112, 113). However, a complete picture of this complex process will require further studies to characterize the precise roles (if any) played by such APC subsets in the induction of central tolerance to peripherally-acquired Ag during fetal development. Regardless, the collective outcome of these processes is a functional T cell repertoire educated for tolerance to self-Ag and NIMA

Experimental support for central tolerance induction to peripherally-acquired antigen

Looking specifically at using PNTx to induce immune tolerance to a circulating protein like FVIII, experimental studies performed in post-natal mice have identified the steady-state myeloid conventional DC2 (cDC2) and plasmacytoid DC (pDC) subsets of APC as being responsible for presenting peripherally-acquired antigens within the thymus (109, 114–117). The first of these studies demonstrated that cDC2 capture bloodstream Ag, endocytose the Ag in a clathrin-dependent manner, migrate to the thymus via C-C chemokine receptor type 2 (CCR2), and finally present the Ag at the cortical parenchyma (118, 119). Subsequent studies affirmed that cDC2-presented Ag can contribute to central tolerance by inducing thymocyte recruitment to the medulla and driving clonal deletion of Ag-reactive thymocytes, as well as inducing the differentiation of Ag-specific nT_reg (120–123). More recent studies suggest that cDC2 (124, 125), participate in Ag-transfer via MHC-I and MHC-II uptake from medullary TECs (126).

A variety of biomarkers have been proposed to identify subsets of pDC that have tolerogenic properties. In early studies, resting CCR9⁺ pDC were identified to have immunosuppressive properties when they were shown to induce the differentiation of Ag-specific T_reg cells to peripherally-acquired Ag (127, 128). Soon after, Martin-Gayo et al. (129) provided compelling evidence that CD13⁻ pDC present within the human thymus are pro-tolerogenic by showing that these cells were able to induce the differentiation of positively-selected DP thymocytes to nT_reg upon CD40L and IL-3 stimulation and CD80/CD86 signaling. In the same year, Hanabuchi et al. showed that, upon stimulation, purified pDC begin to express TSLP receptor, and that upon binding of TEC-produced TSLP, these pDC secrete chemokines that promote medullary recruitment of positively-selected thymocytes (123), while also inducing DP thymocytes to undergo differentiation to nT_reg (130). The authors also noted that the tolerogenic pDC secreted higher levels of IL-10, compared to that secreted by conventional DC.

In addition to cDC2 tolerogenic effect within the thymus, fetal cDC2 can also induce peripheral tolerance to alloantigen by localizing to lymph nodes and mediating antigen-specific inducible T_reg cell differentiation (93).

Experimental Work Optimizing PNTx in the Context of HA

The therapeutic potential of PNTx has been demonstrated in human patients with primary immunodeficiencies and osteogenesis imperfecta (57, 58, 96, 98, 131). As discussed above, HA can easily and noninvasively be diagnosed prenatally. For many medical and biological reasons, PNTx represents an ideal modality for treating HA, and the rationale for this approach to HA treatment has been reviewed previously (87, 88). For PNTx to be a safe and successful treatment approach for HA (and a variety of other genetic diseases as well), it will be critical to use donor cells that can evade HLA-directed immunogenic responses. Almost all clinical prenatal transplants performed to-date have used HSC. When considering the treatment of HA, there is no reason to limit our consideration of potential donor cells to those of hematopoietic origin. Indeed, by expanding the scope of putative donor cells to include those that possess a phenotype similar to that of the fetal trophoblast or placenta, it may be far easier to induce tolerance to donor cells, due to their lack of expression of immunogenic HLA Class II antigens, and the expression of non-classical Class Ib HLA-G or HLA-E (132). Indeed, there has been a great deal of interest in the use of amniotic fluid- and placenta-derived cells for PNTx to target multiple genetic disorders (133–135).

Since the discovery of stem cells within the amniotic fluid and the subsequent demonstration of their potential as therapeutics (134–136), investigators have been studying mesenchymal stromal cells (MSC) derived from both amniotic fluid (AFSC) and the placenta (PLC) as cellular therapeutics for PNTx. Their relatively non-immunogenic profile and their ability to be extensively expanded without exhibiting genomic instability or undergoing transformation have further fueled these efforts. Looking specifically at their immune-modulating properties, AFSC and PLC have both been shown to inhibit lymphocyte activation, stably produce a tolerogenic secretome, contribute to multiple mechanisms of tissue regeneration, and have demonstrated the potential for long-term engraftment (134, 136–138). In the context of HA treatment, these unique immunological properties may make AFSC or PLC ideal donor cells to provide a non-immunogenic cellular platform for the long-term secretion of therapeutic levels of FVIII, prenatally, and perhaps even postnatally.

When considering the use of PLC as cellular therapeutics, it is important to note that PLC can be derived from both the maternal (decidual tissue) and the fetal side of the placenta. Given their earlier ontogenic state, investigators are actively focused on developing methods to optimize isolation of PLC of fetal origin. Sardesai and colleagues demonstrated that, via a specific cotyledonary core dissection and large-scale explant culture, fetal PLC can be reproducibly isolated (139). Steigman and Fauza also described a method of isolating MSC from the fetal side of the placenta (140) and showed that these cells exhibited a spindle-shaped fibroblast-like appearance characteristic of MSC isolated from adult tissues and had the ability to be extensively expanded ex vivo. MSC isolated in this manner expanded more quickly in vitro than adult MSC and were both less immunogenic and more potent immunosuppressors than their adult counterpart (141–143). More recent studies have further refined these isolation methods by demonstrating that PLC, in similarity to their counterparts in the amniotic fluid, can be further enriched for stem-like properties and broadened differentiative capacity by including a step to select for the c-kit⁺ subpopulation of those cells exhibiting plastic-adherence (144, 145). Despite their promise, however, additional studies are still needed to establish their long-term in vivo stability, differentiative fate, and safety following PNTx. Ideally these cells should have a broad engraftment/integration capability, enabling them to exploit the space and permissive environment created by the rapid fetal growth. In addition, these cells should have low immunogenicity, so they can elude the nascent immune system to engraft at therapeutic levels and persist long enough to induce tolerance.

Experimental Work Optimizing in Utero Gene Therapy (IUGT) in the Context of HA

Given the extremely low endogenous level of FVIII production/secretion by most cell types being considered as therapeutics (146, 147), it is unlikely that PNTx with a single infusion of cells that endogenously produce and secrete FVIII would be sufficient to achieve normal circulating FVIII levels (100–200ng/mL). Rather, to achieve clinically meaningful levels of circulating FVIII following PNTx, it will likely be necessary to either inject vectors directly into the fetal recipient (as is being done postnatally in ongoing AAV-based clinical trials) or infuse cells that have been modified to constitutively and robustly express native or expression-enhanced FVIII transgenes.

While gene delivery technology has advanced exponentially over the past decades, FVIII is a challenging protein to express ectopically, as it is large and heavily glycosylated, which places a great deal of stress on the cell’s secretory machinery (148–152). To address this issue, investigators have pursued multiple avenues in efforts to maximize the amount of exogenous vector-derived FVIII that reaches the circulation. These efforts have included modifying native gene promoters and/or synthesizing artificial promoters, performing codon-optimization of the FVIII transgene, and bioengineering FVIII to alter its protein structure, to improve transcription, translation, and secretion, respectively. In a striking example of the first of these approaches, Brown and colleagues developed a bioengineered FVIII cassette, delivered it in the context of an AAV vector, and demonstrated its potent therapeutic potential in the management of HA (153). Compared to the strongest known liver-directed promoter, the 252-bp HLP (154), the authors iteratively designed and developed a shorter 146-bp synthetic promoter comprised of a transcription-driving α-microglobulin/Bikunin precursor (abp) fragment stripped of non-transcription factor-binding regions, a Xenopus laevis albumin (SynO) fragment, and a transcription start site (TSS). The authors termed this new engineered promoter the hepatic combinatorial bundle (HCB), and they showed that it is capable of driving 14-fold higher FVIII expression (vs. HLP), as measured by FVIII chromogenic assay, after hydrodynamic injection of plasmid DNA in a murine HA model.

These same authors also described a method for codon-optimization to maximize expression within the desired target tissue and showed that, in the context of HA, a liver codon optimization can yield 7-fold higher FVIII expression in vivo compared to traditional human genome coding DNA-based codon optimization methods derived from an ancestral FVIII-encoding gene. Very intriguingly, a vector dose of 1×10¹¹ vector genomes (vg)/kg of a liver codon-optimized AAV-FVIII vector with a porcine A2 domain-encoding region was shown to sustain FVIII:C greater than 100 IU/dL after 16 weeks in murine HA models. This is a 600-fold lower dosage in comparison to a recent clinical trial that required a single 6×10¹³ vg/kg dose to achieve sustained levels higher than 50 IU/dL (47), attesting to the remarkable gains that can be made in plasma FVIII levels by optimizing the transcription and translation of the FVIII transgene.

Modulation of protein structure has also been shown to improve the efficiency with which FVIII is secreted. Another AAV-based study recently showed that by deleting the furin cleavage site in a human B domain-deleted (hBDD) FVIII cassette, the resultant change in FVIII structure led to a 2- to 4-fold increase in FVIII expression in vivo (155). Collectively, such preclinical evidence strongly argues for the immense promise of vectors containing bioengineered FVIII variants in future GT approaches for HA.

More recently, another research group created an LV construct designed to target FVIII expression to liver sinusoid endothelial cells (which are thought to be the predominant natural site of FVIII synthesis in the body) through the incorporation of a vascular endothelial cadherin (VE-cadherin) promoter in HA mice (156). The study reported sustained FVIII:C levels of 5–6 IU/dL for up to 1-year post-injection in the absence of FVIII inhibitors. This remarkable feat was the result of preventing transgene expression in pDC via the incorporation of a target site for miR-126 (a miRNA found abundantly in pDC) within the LV.

As a result of the advances afforded by this multi-tiered approach to FVIII production and secretion, it is now possible to express an exogenous FVIII cassette in a variety of cells and tissues both in vitro and in vivo, and it has been some of these very advances that enabled the recent postnatal gene therapy clinical trials for the treatment of HA discussed earlier in the review. However, given the uniqueness of the ethical and scientific considerations surrounding the prenatal application of gene therapy, the International Fetal Transplantation and Immunology Society (IFeTIS) recently published a Consensus Statement outlining the key criteria for considering treatment with IUGT to help advance IUGT towards clinical realization. These criteria included: 1) reliable prenatal enzymatic or genetic diagnosis, 2) a strong genotype/phenotype correlation impacting clinical prognosis, and 3) the need to exploit the immunologic environment at the time of treatment to prevent pre- or postnatal immune reaction to the transgene-encoded protein (55, 56). HA meets all of these criteria, highlighting the potential of IUGT as a viable treatment approach for this disease.

Looking first at the approach of directly injecting viral-based vectors into the fetal recipient to mediate gene transfer/correction, a variety of animal model systems and rodent models of human genetic diseases and a wide range of transduction methods have been employed. These studies have collectively demonstrated that direct vector injection-based IUGT can be used to target multiple organs, and phenotypic rescue has been accomplished with this approach in several different disease models (72, 75, 76, 77, 157–203). Our group has spent the last two decades performing IUGT studies in the sheep model (75, 77, 161–166, 193, 197–202), and we have shown that it is possible to take advantage of the unique temporal window of relative immuno-naïveté during early gestation to efficiently deliver exogenous genes to a variety of fetal tissues and induce durable tolerance to the vector-encoded gene product (77, 202). This tolerance induction appears to involve both cellular and humoral mechanisms, since antibody and cellular responses to the transgene product were both significantly diminished in these animals, even several years after IUGT.

Looking specifically at using IUGT to correct the hemophilias, promising studies demonstrated that injection of FIX-encoding lentiviral (LV) or AVV vectors into mouse fetuses resulted in therapeutic levels of FIX and improved coagulation post-IUGT. Furthermore, no immune response developed to FIX, even when the protein was repeatedly injected postnatally (76, 177).

Collectively, the results of all of the afore-mentioned studies strongly imply that IUGT, even if it not curative, would still be an ideal treatment modality for HA, since the induced immune tolerance would ensure that postnatal therapy, be it protein- or gene-based, could proceed safely without any of the immune-related problems that currently plague HA treatment.

In a more recent example of the potential of IUGT, Peranteau and colleagues performed direct injection AAV-mediated IUGT in fetal sheep at 60 to 65 days of gestation (term = 150 days). These studies showed sustained expression of AAV vector-encoded GFP for up to 6 months postnatally, as well as the induction of immune tolerance to the vector-encoded GFP (72). Similarly, another study by Chan and colleagues achieved long-term expression and immune tolerance to human coagulation FIX for up to 6 years after a single, direct in utero injection of an AAV-FIX vector to healthy fetal nonhuman primates, thereby establishing proof-of-concept for treating hemophilia B prior to birth in this highly translational animal model (203).

Interestingly, however, in both of these studies (performed in two different species), the authors observed postnatal immunity to the capsid of the specific serotype of AAV used in the study. The absence of immune tolerance to the vector capsid is perhaps not surprising, given the very brief time that the capsid proteins/peptides would be expected to be present following injection. Given the brevity of exposure of the fetal immune system to the capsid and its proteasomal degradation products, however, the induction of immunity is unexpected, and such immunity would preclude the ability to perform a subsequent postnatal “boost” with the same vector preparation to enhance the levels of transgene expression, if needed to reach a therapeutic target range.

While new AAV serotypes and AAV hybridization approaches promise to one day enable avoidance of such AAV capsid-directed immunity (204–206), to-date the prevalence of pre-existing anti-AAV antibodies in the general population and the possibility of acquired anti-AAV capsid immunity has thwarted AAV-based clinical approaches to treat multiple genetic diseases. These immunological hurdles, combined with significant safety concerns over off-target effects (87, 161, 163) and the possible toxicity of delivering viral vectors directly in vivo (207), has fueled the search for alternative approaches for safely delivering a FVIII transgene to correct HA.

One of the most promising of these approaches is using cells as vehicles to deliver a FVIII expression cassette. In support of this approach, a number of preclinical studies have established the feasibility of using a variety of viral-based vectors encoding FVIII cassettes to achieve high levels of secreted FVIII:C activity in vitro and in vivo from a variety of clinically relevant cell types (153, 155, 208–221). Among the candidate viral vectors, those based upon lentiviruses have shown great promise, due to their ability to transduce a wide range of cell types, to permanently integrate their genetic cargo into the genome of the host, and to efficiently transduce both dividing and quiescent target cells (222–224). It is important to note that LV vectors are currently being used in multiple clinical trials [ClinicalTrials.gov and (205)], establishing both their safety and their tremendous utility/potential. Looking specifically at using LV vectors in the context of HA, experimental studies by Doering and colleagues using a self-inactivating (SIN) LV vector have shown promising results in postnatal GT for HA (209). In this study, the authors incorporated a bioengineered hybrid FVIII gene containing the porcine A2-domain to achieve enhanced secretion, and they utilized the CD68 promoter to drive myeloid-specific expression. In combination, these two modifications led to a marked enhancement in the circulating levels of FVIII:C in HA mice following transplantation with LV-transduced human CD34⁺ HSC.

Shi and colleagues have spent several years exploring the possibility of harnessing the natural potential of platelets as transporters of FVIII in the context of gene therapy treatments for HA. In their first study, they showed that by combining the use of an LV vector (2bF8LV) encoding a platelet-specific promoter (αIIb) driving FVIII expression, and a drug resistance gene, MGMTP140K, it was possible to transplant these transduced HSC and, after in vivo drug-selection, achieve normalized FVIII:C in the absence of anti-FVIII antibodies for up to 12 weeks post-GT in a murine HA model (225, 226). Miao and colleagues greatly simplified this complex and potentially toxic approach by showing that it is possible to administer a single intraosseous injection of a FVIII-encoding LV vector and thereby transduce HSC in vivo and achieve platelet-specific expression of FVIII that corrects the bleeding phenotype of HA mice, even in the presence of inhibitors (227). More recently, Shi and colleagues demonstrated that, using the same 2bF8LV GT, peripheral tolerance may be occurring via increased T_reg cell levels upon FVIII re-stimulation (228). Notably in this study (228), however, successful peripheral tolerance was only achieved in conjunction with either irradiation or busulfan + antithymocyte globulin dosages. In further mechanistic studies, Shi and colleagues used an ovalbumin transgene to show that platelets store and transport this vector-encoded protein in α granules post-GT, and that this platelet-targeted approach to GT results in ovalbumin-specific humoral and cellular peripheral tolerance in vivo (229).

In contrast to the limited peripheral tolerance that has been achieved in the preceding postnatal studies, IUGT may offer the potential to induce “permanent” central tolerance to FVIII, without the need for nonspecific immunomodulation regimens, via deletion of reactive FVIII-specific T cell deletion or the induction of natural T_reg differentiation during development of the fetal lymphocyte repertoire. Nevertheless, studies are still needed to define the minimal circulating FVIII levels that are required following IUGT for APC (potentially cDC2 and/or CD123⁺ pDC) to effectively uptake and present FVIII peptide at an adequate affinity and avidity to lymphocyte progenitors (thymocytes) during fetal development to induce central tolerance (Figure 1).

Figure 1. Intrathymic presentation of peripherally-acquired FVIII by APC during fetal development.

a) After PNTx/IUGT, vector-encoded FVIII produced by transplanted cells, or by vector-modified cells, is taken up by antigen-presenting cells (APC) either in peripheral circulation or within the fetal thymus by resident APC. b) The FVIII peptide is then presented by MHC molecules on fetal APC to T cell progenitors (thymocytes) in the thymic medulla. Throughout gestation, thymocytes with low/moderate affinity will exit the thymus as naïve T cells that are tolerant to the vector-encoded FVIII, while those with modest-high affinity will become FVIII-specific natural T_reg cells, and those with excessively high affinity will undergo cell death, preventing postnatal anti-FVIII immunity. TEC = thymic epithelial cell.

To date, there has been only a single study looking specifically at the possibility of using gene-modified cells to treat HA via IUGT. This very recent report demonstrated the feasibility of transducing human PLC with a LV encoding a truncated FVIII transgene for use in PNTx (230). The transduced PLC secreted biologically active FVIII coagulation activity (FVIII:C) at levels of 1.11 IU/10⁶ cells and maintained the capacity for trilineage differentiation post-transduction. The authors then performed PNTx with these transduced PMSC in healthy wild-type mice and demonstrated the presence of donor cells in several organs soon after birth. While establishing basic proof-of-principle, the authors did not assess long-term engraftment of the PMSC, the circulating levels of vector-derived FVIII, nor the potential for PNTx to induce immune tolerance to FVIII, leaving unanswered the question of the long-term therapeutic potential of this approach.

A new exciting gene transfer technology that warrants mention is that of non-viral delivery platforms, which are becoming more prevalent in preclinical studies. A striking example of the remarkable advances being made in this field is the recent innovative study by Saltzman and colleagues, who performed direct intravenous and intra-amniotic injections in utero of poly(lactic-co-glycolic acid)-encapsulated nanoparticles (NP) containing triplex-forming peptide nucleic acids (PNA) to correct the IVS2–654 mutation that causes β-thalassemia (231). In contrast to the viral vectors used in most IUGT studies to-date, this biodegradable and biocompatible NP-based system uses high-fidelity endogenous DNA repair mechanisms, including homology-directed repair pathways, thereby minimizing the possibility of insertional mutagenesis. This approach yielded efficient delivery to the fetal liver and led to significant phenotypic improvement of β-thalassemia postnatally. Equally as important as its therapeutic efficacy, this approach did not result in any developmental abnormalities nor inflammatory cytokine responses, and the authors documented an impressive 100% survival rate at 500 days post-IUGT. The authors insightfully capitalized on the naturally-occurring high gene-editing frequency in the fetal liver and fetal BM (232) to drive site-directed donor gene integration and confirmed that higher gene-editing levels are achieved in utero when compared to postnatally, and thereby highlighting another unique therapeutic advantage of IUGT (233).

Placing this new non-viral delivery system in the context of HA, one can envision future studies using PNA-containing NP to achieve site-specific integration of expression/secretion-optimized engineered FVIII transgenes within AFSC and/or PLC which could then be used to treat HA via PNTx and bypass possible risks/concerns associated with the use of viral-based vectors in the fetal recipient. Such studies would obviously need to include optimization of the dosing and gene-editing efficiency of the PNA-containing NP in utero, validation of specificity of genomic integration of the donor gene, and assessment of the long-term stability of the gene-corrected cells in vitro and in vivo in preclinical animal models of HA. And most importantly, these studies will need to rigorously verify the safety of this approach to IUGT.

Concluding remarks

Great strides have been made in the understanding of the structure and function of FVIII, the deficiency of which leads to Hemophilia A (HA), and this improved understanding has enabled vast improvements in the standard of care for children and adults suffering from the disease. The identification and characterization of the human FVIII gene by Jane Gitschier in 1984 (1), followed by the concept of B domain deletion in 1988, which was pioneered by the John Toole laboratory, as well as the characterization of FVIII activation requirements by Debra Pittman in the same year, each helped to pave the way for the clinical reality of prophylactic FVIII replacement therapy that is available today.

Currently, prophylactic FVIII infusions safely and effectively enable sustained restoration of normal hemostasis, thereby preventing bleeding events and enabling HA patients to live near normal lifespans; both remarkable clinical achievements. Unfortunately, this standard of care is not accessible to many of the world’s HA patients nor is it financially palatable. Current treatment regimens are lifelong and noncurative, leading to a reduced quality-of-life, and are estimated to cost over $300,000 per year, with much higher costs for the >30% of severe HA patients who develop FVIII inhibitors (37, 38). Clearly, there is much room for improvement in the clinical management of HA. In contrast to current replacement therapy, the successful delivery of FVIII-encoding vectors or cells modified with such a vector could promise prenatal correction of HA and long-lasting, ideally lifelong, phenotypic correction of the disease. Equally as compelling, intervening early in gestation should enable the induction of central tolerance to the exogenous FVIII protein by piggybacking upon the naturally-occurring intrathymic, peripheral antigen presentation via fetal APC subsets throughout prenatal lymphocyte selection (Figure 1). Thus, even if prenatal treatment did not provide long-term curative circulating levels of FVIII, it should ensure the patient would never form inhibitors to FVIII, thus overcoming the major complication in the management of HA patients today.

Regarding ensuring adequate levels of engraftment of donor cells following PNTx, the documented tolerogenic propensity of fetal immune cells to maternally-derived haploidentical cells supports the potential for using maternal cells for PNTx to minimize the risks of rejection/graft failure. Looking beyond the HSC that have been used in the majority of clinical PNTx performed to-date, many experimental studies have also demonstrated meaningful levels of engraftment following PNTx using mesenchymal stromal cells derived from chorionic villous sampling, amniotic fluid, full-term placenta, or the umbilical cord, all of which possess intrinsic advantages over HSC with respect to immunogenicity, genomic stability, and the ability to be expanded extensively in vitro (133, 230).

To maximize FVIII biosynthesis following gene delivery, elegant studies have demonstrated the substantial impact that target cell-specific promoters, codon optimization, and bioengineering of the FVIII protein, as well as subtle protein structural alterations to bypass the unfolded protein response pathways, can all have on the levels of FVIII that are produced and secreted into the circulation.

Combining these advances in FVIII biology with those in PNTx, the field is uniquely poised to begin developing and optimizing prenatal treatments for HA. It is important to note that the great strides made in recent years with postnatal GT are the result of decades of extensive experimental work to define the efficacy and safety of the approaches to be applied to human patients and to move through the extensive regulatory intricacies of this new, untested molecular therapy. Based on the tremendous advances that have been made in recent years, dual initiatives have been made by the NIH and FDA to better streamline the clinical research oversight system as it pertains to GT (234), including changes to the review and reporting processes to encourage and ensure full transparency with respect to the safety and efficacy of candidate therapies.

With these new streamlined processes and the remarkable ongoing advances in both non-viral and viral gene delivery, we are optimistic that IUGT will soon become a clinically viable treatment option for patients with HA (and numerous other genetic disorders, as well). Prior to clinical implementation however, critical questions concerning IUGT’s long-term efficacy, safety (to the fetus and the mother), and its potential to induce immunological tolerance to the vector-encoded proteins must first be answered in biologically relevant and clinically-predictive animal models.