Hemophilia “A” gene therapy: Lost in translation

Successful gene therapy for hemophilia A had been considered as one of the “holy grails” in the gene therapy field. An important milestone toward realizing this goal had been achieved (https://www.ema.europa.eu/en/news/first-gene-therapy-treat-severe-haemophilia) following the first conditional marketing authorization approval in the European Union for an adeno-associated virus type 5 (AAV5)-based gene therapy strategy.¹ Most of the treated patients remained off prophylaxis and bleed-free 2 years after dosing and up to 5 years in a supporting trial.^2,3 However, circulating levels of the therapeutic factor VIII (FVIII) protein gradually declined, consistent with similar dynamics of FVIII expression in the majority of patients in other gene therapy trials for hemophilia A.⁴ The underlying mechanism for these unstable FVIII levels are not fully understood. In this issue of Molecular Therapy, Butterfield and colleagues now show, in comprehensive preclinical studies in hemophilia A mice, that this loss in FVIII expression may be due—at least in part—to translational shutdown of FVIII protein expression.⁵ Interestingly, this decline in FVIII expression could be rescued by blocking interleukin-15 (IL-15)-dependent natural killer (NK) and CD8⁺ T cell responses, suggesting possible avenues to stabilize FVIII levels upon clinical translation.

Hemophilia A and B are severe bleeding disorders caused by deficiencies in FVIII and FIX, respectively. Patients with hemophilia can be treated successfully by protein-substitution therapy (PST) with plasma-derived or recombinant FVIII or FIX proteins. However, PST does not constitute a cure, and clotting factor levels decline to subtherapeutic levels after a few days. This could be prolonged based on the use of bio-engineered clotting factors with extended half-life (EHL). Nevertheless, even with EHL products, maintaining circulating factor levels above the therapeutic threshold remained a challenge, mandating frequent and life-long FVIII or FIX retreatment to prevent potential life-threatening bleeds that hang as a proverbial sword of Damocles over the patients’ heads.

Since the early days of the gene therapy field, more than 30 years ago, hemophilia had been considered an ideal target disease for gene therapy that could potentially offer a long-term solution to prevent such life-threatening bleeds and alleviate the complications of inadequate factor levels. Preclinical studies in hemophilic mice and dog models, complemented with studies in non-human primates, showed long-term safety and efficacy after gene therapy, offering real hope that successful clinical translation was imminent. Relatively stable multi-year FIX levels have been attained after AAV-based gene therapy in patients with hemophilia B, though this typically required transient immune suppression with corticosteroids to prevent immune clearance of AAV-transduced hepatocytes and to reduce liver inflammation and immunotoxicity (recently reviewed in VandenDriessche et al.⁶). Therapeutic FVIII levels could also be attained after AAV gene therapy in patients with hemophilia A, consistent with diminished FVIII usage and reduced bleeds.^2,3,4 This prompted conditional regulatory approval of valoctocogene roxaparvovec, an AAV5-based gene therapy product that relies on the expression of a codon-optimized B domain-deleted FVIII transgene expressed from a hepatocyte-specific promoter (https://www.ema.europa.eu/en/news/first-gene-therapy-treat-severe-haemophilia).¹ The outcomes of the hemophilia B gene therapy trials created the expectation that attaining multi-year stable FVIII expression levels after AAV gene therapy in patients with hemophilia A was also within reach. Surprisingly, however, this turned out not to be the case. Contrary to the translational experience with AAV-FIX, FVIII expression appeared unstable and typically declined over the course of several years post-dosing despite immune suppression with corticosteroids to block liver inflammation. Moreover, there was significant interpatient variability in the FVIII levels. Several immune and non-immune hypotheses have been proposed, but the exact mechanism of this FVIII decline remained elusive. This was compounded by the absence of a preclinical animal model that adequately recapitulates these unexpected FVIII expression dynamics observed in clinical trials.

In this issue, Butterfield and colleagues now assessed the possible underlying mechanisms of this gradual loss of FVIII expression based on comprehensive animal studies.⁵ An alternative preclinical FVIII-deficient hemophilic animal model was therefore developed in which the hemophilia-inducing mutation in the F8 gene was present on a BALB/c genetic background instead of the conventional C57Bl6/129 background. These animals were injected with an AAV8 vector expressing a B domain-deleted FVIII (i.e., AAV8-FVIIIΔB) from a hepatocyte-specific promoter and were subsequently given transient immune suppression with rapamycin to prevent antibody formation against FVIII. Despite the lack of anti-FVIII antibody formation, FVIII expression gradually declined over the course of 4 months after AAV8-FVIIIΔB gene therapy, mimicking the dynamics of FVIII expression observed in the hemophilia A gene therapy clinical trials. Remarkably, FVIII expression declined despite persistence of the therapeutic FVIII transgene and FVIII mRNA, suggesting a translational block rather than loss of transduced hepatocytes or transcriptional repression. Most importantly, FVIII protein production could be restored, at least partially, by blocking IL-15 signaling. This could be achieved by targeting either the IL-15 cytokine itself or the CD122 subunit of the cognate IL-15 receptor. Blocking IL-15 signaling suppressed CD8⁺ T and NK cell responses but without affecting FVIII mRNA levels or FVIII gene copy numbers.

Taken together, this reveals unexpected new mechanistic insights pertaining to a possible role of the immune system, involving CD8⁺ T cells and NK cells, in shutting down FVIII protein production without actually eliminating the transduced hepatocytes through conventional cytolytic mechanisms. The exact molecular mechanism of this translational shutdown of FVIII expression is not fully understood. FVIII is normally expressed by endothelial cells and not by hepatocytes. Forcing ectopic FVIII expression in hepatocytes may have triggered an endoplasmic reticulum (ER) cellular stress response that, in turn, may have contributed to the loss of FVIII translation as a possible cellular survival mechanism. The observed increased phosphorylation of the translational elongation factor eIF2α is consistent with such a cellular stress response, which may not only have contributed to the translational shutdown of FVIII but perhaps also triggered global changes in protein translation, but this awaits further experimental confirmation. Since the immune response is known to influence cellular stress, and vice versa, this complex interplay may have played a key role in controlling the translational shutdown of FVIII. Consequently, blocking IL-15R-mediated signaling could have interrupted this interplay between the immune response and cellular stress, promoting the persistence of FVIII protein production. Though the present data support that FVIII expression is lost at the translational level, recent evidence suggests that durability of transgene expression likely also depends on the vector manufacturing method and the size of the vector genome.⁷ In particular, the relatively large size of the FVIIIΔB cDNA inevitably results in over-sized AAV genomes (∼5 kb). When produced using the baculovirus/Sf insect cell-based platform, such over-sized AAV genomes were prone to epigenetic regulation that reduced vector genome accessibility. This was consistent with a reduction in active histone marks (H3K27ac/H3K4me3) and decreased RNA transcription. Though these studies were based on another transgene (α1-antitrypsin as opposed to FVIIIΔB), the results suggest that the decline in FVIII expression observed in some of the gene therapy clinical trials for hemophilia A could have been due, at least in part, to such epigenetic effects.

Translating gene therapies from bench to bedside remains challenging, and hemophilia A gene therapy is no exception. Modeling the complex interactions between the gene therapy vector, the therapeutic protein, the transduced cells, and the immune system in preclinical animal models is not straightforward. Despite these challenges, the study by Butterfield and colleagues demonstrate that it is possible to experimentally interrogate the durability and variability of FVIII expression in a carefully designed animal model.⁵ This led to the discovery of an unexpected novel mechanism that controls FVIII expression and possibly paves the way toward more effective and targeted immune control strategies. Understanding the underlying mechanisms that contribute to this loss of FVIII expression at the level of protein translation may maximize our chances of ensuring that gene therapy is ultimately not lost in translation when expanding the treatment options of patients with severe hemophilia A.