Platelet-targeted hyperfunctional FIX gene therapy for hemophilia B mice even with preexisting anti-FIX immunity

Hemophilia B (HB) is a genetic bleeding disorder resulting from a factor IX (FIX) deficiency.¹  Protein replacement therapy is effective for the disease, but it is constrained by the short half-life of FIX, requiring frequent infusions.^2-6  Furthermore, 5% of patients will develop neutralizing antibodies (inhibitors) against FIX,^7,8  for which there is no effective approach for inducing immune tolerance.⁹  Moreover, anaphylactic reaction to the infused FIX protein in patients with inhibitors is a daunting problem that increases the risk of morbidity and mortality.^7,10-14  Therefore, an effective protocol for treating patients with inhibitors is urgently needed.

Gene therapy is an alternative for HB treatment. Substantial progress in preclinical studies has been achieved in the last 2 decades.^15-36  It has been shown that lentivirus (LV)- or adeno-associated virus (AAV)–mediated liver-targeted F9 gene transfer can reverse preexisting anti-FIX immunity and subsequently establish therapeutic levels of FIX in HB animal models,^15,32  but 25% of inhibitor-prone mice were nonresponders with no FIX detectable after treatment.¹⁵  Clinical trials involving HB patients show that infusion of the AAV8 vector encoding codon-optimized FIX driven by a liver-specific promoter leads to sustained therapeutic levels of FIX expression.^37-40  Furthermore, a combined effect of codon optimization and the gain-of-function FIX-Padua variant (R338L) can significantly enhance the efficacy of liver-targeted gene therapy in HB.^41,42  These data are very encouraging, but an AAV-mediated liver-targeted protocol can be applied only to adults without liver disease or anti-AAV antibodies, which are present in 30% to 50% of the population.^43-45  Thus, an alternative gene therapy approach is desired.

We have developed a platelet-specific gene therapy protocol for hemophiliacs that targets transgene expression to platelets under the control of the platelet-specific αIIb promoter.^46-53  We have shown that platelet-specific FVIII expression (2bF8) can restore hemostasis in hemophilia A (HA) mice, even those with inhibitors.^47,49,52  But platelet-specific FIX expression rescues bleeding diathesis only in HB mice without inhibitors⁵⁴  because FIX does not have a protective carrier protein, unlike FVIII which is protected by von Willebrand factor (VWF).^55-57  However, platelet-FIX gene therapy can induce FIX-specific immune tolerance in HB mice in the noninhibitor model.⁵³  Here we explored platelet-targeted codon-optimized hyperfunctional FIX gene therapy for HB, even in mice with preexisting anti-FIX immunity.

The following paragraphs briefly summarize the more detailed descriptions provided in the supplemental Data regarding antibodies and reagents, as well as methods and statistical analyses used in this study. FIX-deficient (FIX^null) mice in either a C57BL/6 background (Model 1) or in a B6-129S mixed background (Model 2) were used. The construct pWPT-2bF9 with wild-type human FIX (WT-hFIX) driven by the αIIb promoter was created as reported.⁵⁴  The novel lentiviral vector, pWPT-2bCoF9R338L harboring a codon-optimized hFIX-Padua^58,59  (CoF9R338L) directed by the αIIb promoter was constructed by replacing WT-hFIX in pWPT-2bF9 with CoF9R338L. 2bCoF9R338L and 2bF9 lentiviruses (LVs) were produced as previously reported.^48,60,61 

Sca-1⁺ cells isolated from FIX^null mice were transduced with 2bCoF9R338L or 2bF9 LV following the procedures previously described.⁵³  Transduced cells were transplanted into each recipient preconditioned with either sublethal 6.6 Gy or lethal 11 Gy total body irradiation (TBI). For the inhibitor model, both donor and recipient FIX^null mice were preimmunized with recombinant hFIX (rhFIX) plus incomplete Freund’s adjuvant (IFA) to induce anti-FIX antibody development.

The 2bCoF9R338L transgene expression was determined by polymerase chain reaction (PCR) and quantitative real-time PCR (qRT-PCR) analysis. FIX expression levels were determined by a chromogenic assay for FIX activity (FIX:C) and by enzyme-linked immunosorbent assay (ELISA) for FIX antigen (FIX:Ag) on platelet lysates as previously described.⁵⁴  To examine whether platelet-FIX is γ-carboxylated, we performed a barium sulfate (BaSO₄) precipitation assay.⁶²  Intracellular location of the neoprotein expression in 2bCoF9R338L-transduced platelets was determined by confocal microscopy, and the percentage of platelets that expressed hFIX in recipients was determined by fluorescence-activated cell sorting (FACS) as reported.⁵³  The hemophilic bleeding phenotype in 2bCoF9R338L-transduced recipients was assessed by tail bleeding test, joint injury, ROTEM, and native whole-blood thrombin generation assay (nWB-TGA) as reported.^46,54,63-66 

Anti-FIX inhibitor titers were quantified by Bethesda assay, and anti-FIX total immunoglobulin G (IgG) titers were determined by ELISA as reported.⁵⁴  For the noninhibitor model, 8 months after hematopoietic stem cell (HSC) transplantation (HSCT), recipients were challenged with rhFIX plus IFA to determine whether immune tolerance was established after 2bCoF9R338L gene therapy. For the inhibitor model, when inhibitor titers dropped to undetectable levels, recipients were re-challenged with rhFIX by IV administration to determine whether immune tolerance was developed after 2bCoF9R338L gene therapy.

Two strains of FIX^null mice, in either a C57BL/6 (Model 1) or a B6-129S (Model 2) background, were used to evaluate the efficacy of 2bCoF9R338L gene therapy (Figure 1A). 2bCoF9R338L provirus DNA was detected by PCR in all 2bCoF9R338L-transduced recipients (supplemental Figure 1). In Model 1, platelet-FIX:C and FIX:Ag levels in 2bCoF9R338L-transduced recipients were 61.29 ± 10.24 and 12.85 ± 5.68 mU/10⁸ platelets at week 4, but they significantly decreased to 28.58 ± 14.5 and 4.38 ± 1.79 mU/10⁸ platelets, respectively, at week 13 after HSCT (Figure 1B). The ratio of platelet-FIX:C to FIX:Ag was 5.16 to 6.38 (Figure 1C). In contrast, platelet-FIX expression in Model 2 was sustained in all transduced recipients throughout the study period, with platelet-FIX:C and FIX:Ag from 65.59 ± 10.66 and 9.77 ± 3.05 mU/10⁸ platelets at week 5 to 66.38 ± 15.55 and 11.24 ± 2.69 mU/10⁸ platelets, respectively, at week 57 after HSCT (Figure 1D). The ratio of platelet-FIX:C to FIX:Ag was 6.14 to 7.89 (Figure 1E).

The average platelet-FIX:C in Model 1 was significantly lower than that in Model 2 (Figure 1F), which was significantly higher than FIX expression in 2bF9-transduced recipients (supplemental Figure 2A). There were no significant differences in the average ratio of platelet-FIX:C to FIX:Ag or the average copy number of 2bCoF9R338L proviral DNA per cell between Model 1 and Model 2 (Figure 1G-H).

To further confirm whether long-term platelet-FIX expression was achieved after gene therapy, sequential transplantations were carried out from Model 2. Bone marrow (BM) cells collected from the primary (1°) recipients were transplanted into secondary (2°) and then tertiary (3°) recipients. The platelet-FIX:C expression levels in the sequential transplantation recipients were comparable to those in the primary recipients (Figure 1I).

To examine the transduction efficiency of 2bCoF9R338L gene therapy in FIX^null mice, we used FACS to analyze hFIX expression in platelets. The percentage of hFIX⁺ platelets in Model 2 after 2bCoF9R338L gene therapy was significantly higher than that in Model 1 (Figure 2A-B). The absolute levels of FIX:Ag and FIX:C in transduced platelets in the 2 models were comparable (Figure 2C) when calculated according to hFIX⁺ platelets measured by FACS. The percentages of hFIX⁺ platelets were similar between the 2bF9 and 2bCoF9R338L groups using Model 1 (supplemental Figure 2B).

To investigate the immune response in 2bCoF9R338L-transduced recipients, we monitored anti-FIX inhibitors after gene therapy, and no inhibitors were detected in transduced recipients. When recipients were challenged with rhFIX plus IFA, no Model 1 mice developed inhibitors; 3 of 7 Model 2 recipients developed low-titer inhibitors (1.1 to 2 Bethesda units per mL [BU/mL]), which subsequently disappeared within 4 weeks. The incidences of inhibitor development and the titers in transduced groups were significantly lower than in the control FIX^null group, in which all mice developed inhibitors with titers of 54.43 ± 47.54 BU/mL when the same immunization protocol was used (Figure 2D-E).

To assess the hemophilic phenotype in 2bCoF9R338L-transduced recipients, we used both in vivo models and ex vivo assays. In the 6-hour tail bleeding test, none of the FIX^null mice stopped bleeding within 6 hours, but all 2bCoF9R338L-transduced recipients did, with bleeding times of 2.5 ± 1.29 hours in Model 1 and 1.43 ± 0.79 hours in Model 2, which was not different compared with the WT control (Figure 3A). After the test, the percentages of remaining hemoglobin in 2bCoF9R338L-transduced recipients were significantly higher than those in FIX^null mice but not different compared with the WT control (Figure 3B).

When animals were challenged with a needle-induced joint injury, the diameter of injured joints (right knee) in FIX^null mice increased 1.159-fold ± 0.059-fold 48 hours after injury, which was significantly larger compared with the uninjured left knee control (0.957 ± 0.005-fold) (Figure 3C). Notably, in 2bCoF9R338L-transduced recipients, there was no significant change in the diameter of the injured joint after injury, and the ratio of diameter after the injury to the diameter at baseline was comparable to that in the uninjured intra-animal control (Figure 3C), demonstrating that platelet-CoF9R338L expression can effectively prevent joint bleeding.

The hemostatic efficacy of 2bCoF9R338L gene therapy was further assessed by ROTEM and nWB-TGA analysis. As shown in Figure 3D, all parameters in ROTEM, including whole-blood clotting time (CT), clot formation time (CFT), maximum clot firmness (MCF), and α-angle, in the 2bCoF9R338L group were significantly improved compared with the FIX^null group. There were no differences in CFT and MCF between the 2bCoF9R338L and WT groups, but CT was longer and α-angle was lower in the 2bCoF9R338L group compared with the WT group. When nWB-TGA was conducted, all parameters, including lag time (LT), peak time (PT), peak thrombin (PTh), and endogenous thrombin potential (ETP), in the 2bCoF9R338L group were also significantly improved compared with the FIX^null group (Figure 3E). There was no difference in PTh and ETP between the 2bCoF9R338L and WT groups, but LT and PT were longer in the 2bCoF9R338L group compared with the WT group (Figure 3E). Taken together, these results demonstrate that targeting codon-optimized FIX-Padua expression by lentiviral gene delivery to HSCs can introduce sustained hyperfunctional FIX expression in platelets, resulting in phenotypic correction and immune tolerance induction in HB mice.

To investigate the feasibility of 2bCoF9R338L gene therapy for the inhibitor model, we used B6-129S FIX^null mice. 2bCoF9R338L-transduced cells were transplanted into rhFIX-primed recipients preconditioned with TBI at either 11 Gy or 6.6 Gy (Figure 4A). After 4 weeks of BM reconstitution, we started to analyze the animals. 2bCoF9R338L proviral DNA was detected by PCR in all transduced recipients (supplemental Figure 1), demonstrating the viability of 2bCoF9R338L genetically modified cells in primed FIX^null mice under either myeloablative or nonmyeloablative preconditioning. The average copy number per cell of 2bCoF9R338L in transduced recipients was similar in the 11-Gy and 6.6-Gy groups (Figure 4B-C). The whole-blood counts in transduced recipients were comparable to those in untransduced transplanted (UnTrans) controls (supplemental Figure 3A-C), demonstrating that platelet-specific FIX-Padua expression does not have an impact on blood cell reconstitution.

The level of platelet-FIX:Ag expression in the 11-Gy group at week 4 was significantly higher than in the 6.6-Gy group, but it was not different at other time points (Figure 4D). Platelet-FIX:C expression levels seem to be lower in both the 11-Gy and 6.6-Gy groups at 4 weeks after gene therapy, but there were no significant differences among time points within each group or between groups (Figure 4E). The average platelet-FIX:Ag and platelet-FIX:C levels were similar in the 11-Gy and 6.6-Gy groups (Figure 4F). The ratios of platelet-FIX:C to FIX:Ag in both groups were comparable, starting at around 2 at week 4 and increasing to 6 at week 16 (Figure 4G).

To confirm whether long-term platelet-FIX expression can be achieved in the setting of preexisting immunity, even under nonmyeloablative preconditioning, sequential transplantation was carried out by transplanting BM cells from primary recipients from the 6.6-Gy group into secondary rhFIX-primed FIX^null mice preconditioned with 6.6 Gy plus bortezomib. Platelet-FIX:C expression levels in the 2° recipients were similar to those obtained from the 1° recipients (Figure 4H) and sustained (Figure 4I) for a total of 2.55 years during our study period (63 weeks for 1° and 70 weeks for 2° recipients).

For comparison, we used 2bF9LV that harbors WT-hFIX to introduce platelet-FIX expression in rhFIX-primed FIX^null mice with either 11 Gy or 6.6 Gy TBI. All recipients had detectable levels of platelet-FIX:C expression at week 4 after 2bF9 gene therapy with levels of 2.31 ± 3.69 and 0.77 ± 0.31 mU/10⁸ platelets, respectively, but 40% in the 11-Gy group and 75% in the 6.6-Gy group dropped to undetectable levels by 18 weeks after HSCT (supplemental Figure 4A-C). Together, these data demonstrate that LV-mediated platelet-specific gene delivery of 2bCoF9R338L can effectively introduce sustained hyperfunctional platelet-FIX expression even in highly primed FIX^null mice.

To examine the intracellular location of hFIX-Padua expression in platelets, we used confocal microscopy. hFIX protein was detected in 2bCoF9R338L-transduced platelets (Figure 5A,D) and colocalized with endogenous VWF protein in platelet α-granules (Figure 5C,F; shown in yellow). Platelets of UnTrans were positive for VWF but negative for hFIX protein as expected (Figure 5G-I). FACS analysis showed that 21.93% ± 11.6% of platelets were positive for hFIX in the 11-Gy group, which was not significantly different compared with the 6.6-Gy group (Figure 5J-K). There were no differences in the absolute hFIX:Ag and hFIX:C expression levels in transduced platelets between the 11-Gy and 6.6-Gy groups when calculated on the basis of the percentage of hFIX⁺carboxylated, we used BaSO₄ to treat platelet lysates from 2bCoF9R338L- and 2bF9-transduced recipients. As shown in Figure 5M, no FIX:Ag remained in the supernatants of BaSO₄-treated platelet lysates from either 2bF9- or 2bCoF9R338L-transduced recipients, in which various levels of platelet-FIX:Ag (0.69-15.62 mU/10⁸ platelets) were added. As expected, nearly 100% of FIX:Ag was precipitated in rhFIX control supernatants after BaSO₄ precipitation (Figure 5N). These data demonstrate that FIX expressed in platelets is γ-carboxylated.

To investigate the immune responses in rhFIX-primed FIX^null mice after 2bCoF9R338L gene therapy, we monitored both anti-FIX inhibitors and total IgG titers in recipients. Inhibitor and total IgG titers declined with time after HSCT in both transduced and untransduced recipients (Figure 6A-B). The half-life of inhibitor disappearance time in the 11-Gy group was significantly shorter than that in the 6.6-Gy and UnTrans control groups (Figure 6C), but there were no differences in anti-FIX total IgG titers among groups (Figure 6D). Importantly, no anaphylaxis occurred in any FIX-primed HB mice after platelet gene therapy.

In 2bCoF9R338L-transduced recipients, the half-lives of anti-FIX inhibitor and total IgG disappearance were negatively correlated with platelet-FIX:Ag and platelet-FIX:C levels (Figure 6E-H). Furthermore, an equal amount of FIX:C in platelet lysates from either 2bF9- or 2bCoF9R338L-transduced recipients could neutralize the same amount of inhibitors, but the same amount of FIX:Ag from 2bCoF9R338L-transduced platelets showed a greater capability to neutralize the anti-FIX inhibitors (supplemental Figure 5). To investigate whether immune tolerance was induced in 2bCoF9R338L-transduced recipients after their inhibitor titers dropped to undetectable levels, animals were re-challenged with rhFIX at 100 U/kg per week by IV administration for 4 weeks. Remarkably, the inhibitor titers remained undetectable after rhFIX rechallenge (Figure 6I). In contrast, all FIX^null mice (n = 4) that were immunized with 2 doses of rhFIX plus IFA and confirmed to have developed inhibitors died within 4 hours when re-challenged with a single dose of rhFIX at 100 U/kg by IV administration (data not shown).

It has been shown that the proteasome inhibitor bortezomib can deplete plasma cells, reducing autoantibody production in autoimmune disease models.^67-70  Here, we evaluated whether bortezomib combined with TBI would facilitate the elimination of anti-FIX antibodies in the inhibitor model. We added bortezomib to 6.6-Gy TBI as preconditioning for 2bCoF9R338L gene therapy in the secondary recipients. We found that inhibitor titers decreased faster in the secondary recipients preconditioned with 6.6 Gy plus bortezomib compared with the primary recipients under 6.6 Gy TBI (Figure 6J). The half-life of anti-FIX total IgG in the secondary recipients was significantly shorter than that in the primary recipients (Figure 6K). As with the primary recipients, once inhibitor titers dropped to undetectable levels in the secondary transplantation recipients, no inhibitors were detected, even after 4 doses of rhFIX rechallenge, demonstrating that immune tolerance was also established in the secondary transplantation recipients. Together, these data demonstrate that platelet-specific CoF9R338L expression does not trigger an anti-FIX memory immune response. Instead, 2bCoF9R338L gene therapy can circumvent anaphylaxis and induce immune tolerance in primed FIX^null mice.

Because the clinical efficacy of platelet-FIX is limited in FIX^null mice with inhibitors,⁵⁴  we did not assess the bleeding phenotype in recipients until inhibitor titers dropped to undetectable levels. When a 6-hour tail bleeding test was used to evaluate the clinical efficacy of 2bCoF9R338L gene therapy in primed mice, the bleeding times in the 11-Gy and 6.6-Gy groups were similar to those in the WT control (Figure 7A). The remaining hemoglobin levels after the test in 2bCoF9R338L-transduced groups were comparable to the WT group but were significantly higher than in the FIX^null control (Figure 7B).

When the needle-induced joint injury model was used to assess the efficacy of 2bCoF9R338L gene therapy in preventing joint bleeds in primed FIX^null mice, the ratio of the injured knee joint’s diameter after injury compared with the baseline was similar to that of the uninjured intra-control knee, which was significantly different compared with the FIX^null control group after injury (Figure 7C). The hemostatic properties in whole blood of 2bCoF9R338L-transduced FIX-primed recipients were further assessed by ROTEM and nWB-TGA. All parameters in both ROTEM and nWB-TGA in 2bCoF9R338L-transduced FIX-primed recipients were significantly improved compared with those in FIX^null controls (Figure 7D-E).

Collectively, these results confirm that platelet-specific 2bCoF9R338L gene therapy can provide therapeutic protein and restore hemostasis once inhibitory antibodies are eradicated in rhFIX-primed FIX^null mice.

The central finding of this study is that lentiviral gene delivery of platelet-specific codon-optimized FIX-Padua expression can effectively introduce sustained hyperfunctional FIX expression in platelets in FIX^null mice even with preexisting immunity without anaphylactic reaction. Platelet-targeted hyperfunctional FIX gene therapy can eradicate anti-FIX inhibitors, induce immune tolerance, and provide therapeutic protein once inhibitors drop to undetectable levels. In addition, we found that using bortezomib as a supplemental agent to TBI preconditioning for 2bCoF9R338L gene therapy can significantly accelerate the eradication of anti-FIX inhibitors in primed FIX^null mice.

Our previous studies demonstrated that 2bF9 gene therapy can restore hemostasis and induce immune tolerance in unprimed FIX^null mice.⁵³  However, platelet-FIX:C levels in transduced mice were only ∼3% in whole blood, even when lethal 11-Gy TBI was used.⁵³  In this study, when CoF9R338L was targeted to platelets, platelet-FIX:C levels in transduced unprimed FIX^null mice reached an average level of 35.08 to 60.77 mU/10⁸ platelets, corresponding to around 70% to 122% in whole blood in WT mice. This is a 26.2-fold to 45.4-fold increase compared with our studies using 2bF9LV under a similar transduction and transplantation protocol. Compared with 2bF9 gene therapy, platelet-FIX:Ag levels in 2bCoF9R338L-transduced recipients were 3.6-fold to 4.6-fold higher, which could be attributed to the effect of codon optimization. The platelet-FIX:C/FIX:Ag ratios in 2bCoF9R338L-transduced recipients were comparable to those in reports targeting FIX-Padua expression to the liver in mice,^36,58  indicating that platelet-derived vs liver-derived FIX has similar functional activity.

FIX is a vitamin K–dependent protein synthesized by hepatocytes in healthy individuals. It requires posttranslational carboxylation to become biologically active. Because γ-carboxylated glutamic acid residues located in the amino-terminal region can efficiently bind to BaSO₄,⁷¹  BaSO₄ precipitation was used to examine the extent of γ-carboxylation of engineered FIX protein.⁷²  Our studies showed that there was no FIX detected in the supernatants of platelet lysates after BaSO₄ precipitation, even in the hyperfunctional FIX-Padua expressers. This confirms that platelet-derived FIX is γ-carboxylated, which demonstrates that megakaryocytes have the capacity to synthesize high levels of hyperfunctional FIX and complete posttranslational carboxylation. The ability of megakaryocytes to synthesize functional, active vitamin K–dependent protein is also evidenced by protein S studies, in which 2.5% of circulating functional protein S is made by megakaryocytes and stored within platelet α-granules.^73-75 

Our studies show that genetic background could affect FIX neoprotein expression in platelet-targeted gene therapy. In Model 1 (a C57BL/6 background), platelet-FIX expression dropped 75% from week 4 to week 24 after 2bCoF9R338L gene therapy, although the levels of platelet-FIX were maintained afterward (data not shown). In contrast, platelet-FIX expression was sustained after gene therapy in Model 2 (a B6-129S mixed background) in both the noninhibitor and inhibitor model studies. The impact of genetic background on platelet-targeted neoprotein expression is also evidenced in platelet-FVIII gene therapy, in which sustained platelet-FVIII expression was maintained in B6-129S FVIII^null mice,^48,52  but it dropped around 54% from week 4 to week 8 in a C57BL/6 background after 2bF8 gene therapy under 6.6-Gy TBI.⁶⁵  Evidence suggests that genetic background influences both humoral and cellular immune responses.^76,77  Because no antibodies against neoprotein were detected in transduced recipients, we speculate that genetic variants between different strains of mice have an impact on the outcome of cellular immune responses in our platelet-specific gene therapy. We know that some CD4 and CD8 T cells are resistant to irradiation, which might lead to an antagonistic process between eliminating the early stages of transduced megakaryocyte progenitors by residual CD8 T cells and inducing immune tolerance by platelet-FIX expression after gene therapy. In the inhibitor model, the platelet-FIX:C levels at the early time points seem to be lower than at later time points, which could be a result of the influence of high-titer inhibitors on the FIX chromogenic assay because any small amount of inhibitory plasma carryover during platelet isolation could neutralize platelet-FIX:C, which would lead to the appearance of lower levels. The other potential reason could be that the FIX-primed immune system has an impact on the uptake of vitamin K, affecting posttranslational carboxylation in the early stages after HSCT.

Gene therapy for HB with preexisting immunity could be challenging because a primed immune system might respond to neoprotein, eliciting a memory immune response or causing anaphylaxis. Furthermore, primed immune cells may kill the transduced cells, leading to gene therapy failure. Our studies show that 2bCoF9R338L gene therapy could prevent anamnestic immune responses and avoid anaphylactic reaction in highly primed FIX^null mice. Using CoF9R338L instead of WT-hFIX significantly improves the functional levels and stability of platelet-FIX expression in platelet-targeted gene therapy in primed FIX^null mice. This could be because 2bCoF9R338L-transduced platelets are more effective than 2bF9-transduced platelets in eradicating or neutralizing anti-FIX inhibitors. Indeed, our results show that there are negative correlations between the levels of platelet-FIX expression and the half-lives of anti-FIX antibody disappearance time in 2bCoF9R338L-transduced recipients.

Although the efficacy of platelet-derived FIX is limited in the presence of anti-FIX inhibitors,⁵⁴  no bleeding episodes occurred in transduced FIX-primed recipients after platelet-targeted gene therapy. All transduced recipients were well-tolerized to ear-tagging after gene therapy, indicating that platelet-FIX-Padua has at least some functional properties, even in the presence of inhibitors. After inhibitors dropped to undetectable, platelet-FIX-Padua could effectively rescue the bleeding phenotype in FIX^null mice in the tail bleeding test and joint injury. The hemostatic efficacy in primed mice was comparable to that in unprimed mice after 2bCoF9R338L gene therapy, demonstrating that the ultimate goal of gene therapy in providing therapeutic protein was achieved. The efficacy and safety issues of FIX-Padua in AAV-mediated liver-targeted gene therapy have been well documented, with remarkable improvement in hemostasis without increasing thrombosis risk in HB mice, dogs, and humans.^33-35,41,58  Our studies show that hemostatic functional properties of platelet-FIX-Padua in FIX^null mice were improved without acceleration of clot formation or thrombin generation compared with WT mice, indicating that platelet-targeted FIX-Padua gene therapy does not increase thrombosis risk in HB mice.

Because the clinical efficacy of platelet-FIX is limited in the presence of inhibitors,⁵⁴  eliminating antibodies is critical, and a regimen that can delete plasma cells is needed. The proteasome inhibitor bortezomib is a promising candidate because it has been shown to effectively eliminate short- and long-lived plasma cells in other autoimmune disease models^67-70,78  and antibodies in patients with HA inhibitors.⁷⁹  In this study, we showed for the first time that the combination of bortezomib and 6.6 Gy TBI as a preconditioning regimen dramatically decreased anti-FIX inhibitors and antibody disappearance time in primed FIX^null mice, and immune tolerance was established after 2bCoF9R338L gene therapy. Our findings strongly suggest that bortezomib could be used as a supplemental agent in a preconditioning regimen to accelerate the elimination of anti-FIX antibodies for platelet-targeted gene therapy for patients with HB who have preexisting immunity.

In summary, we showed that platelet-specific FIX gene therapy can evade anaphylactic reaction and induce immune tolerance in primed HB mice. Using codon-optimized FIX-Padua remarkably enhances platelet-FIX functional levels and provides long-term therapeutic hyperfunctional FIX protein expression in HB mice, even those with preexisting immunity. A combination of bortezomib with 6.6 Gy TBI preconditioning for platelet FIX gene therapy significantly accelerates the elimination of inhibitors in HB mice. Taken together, platelet-specific 2bCoF9R338L gene therapy could be a promising approach for gene therapy of all HB, even in the high-risk setting of preexisting anti-FIX immunity.

This work was supported by a grant from the National Institutes of Health, National Heart, Lung, and Blood Institute (HL-102035) (Q.S.), Bayer Hemophilia Foundation Award (Q.S.), and generous gifts from the Children’s Wisconsin Foundation (Q.S.) and Midwest Athletes Against Childhood Cancer Fund (Q.S.).

Contribution: J.A.S. designed the study, performed experiments, analyzed data, and edited the manuscript; J.C., Y. Chen, Y. Cai, H.Y., and J.G.M. performed experiments and analyzed data; P.E.M. provided critical materials and edited the manuscript; and Q.S. designed the research, supervised the studies, analyzed data, and wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Qizhen Shi, Department of Pediatrics, Medical College of Wisconsin, 8701 Watertown Plank Rd, Milwaukee, WI 53226; e-mail: qshi@versiti.org.