High-Level Expression of Alkaline Phosphatase by Adeno-Associated Virus Vector Ameliorates Pathological Bone Structure in a Hypophosphatasia Mouse Model

Hypophosphatasia (HPP) is a rare congenital disease caused by mutations in the gene encoding tissue-nonspecific alkaline phosphatase (TNALP) [1, 2]. HPP is characterized by defective calcification of hard tissue, pathologic fracture, dyspnea, seizures, and premature tooth loss [3, 4]. It is classified into six different types according to the age at onset and symptoms: perinatal severe HPP, perinatal benign HPP, infantile HPP, childhood HPP, adult HPP, and odonto-HPP [5, 6], with perinatal severe HPP and childhood HPP frequently being fatal [7, 8]. However, treatment for HPP has not been well established.

Whyte et al. first applied enzyme replacement therapy (ERT) by infusion of alkaline phosphatase (ALP)-rich plasma to an infantile patient with HPP [9]. In 2008, Millán et al. developed a novel ERT using bioengineered TNALP. The novel construct was formed by detaching the glycosylphosphatidylinositol (GPI) anchor from TNALP to allow secretion and adding a deca-aspartate motif to the C terminus (TNALP-D₁₀) to increase affinity for bone [10]; the latter was possible because TNALP is an ectoenzyme whose C terminus is anchored to the cell surface by a GPI anchor glycoprotein composed of approximately 500 amino acid residues [11]. This method clearly improved the phenotype of TNALP-deficient mice [10]. Furthermore, this method was applied to patients with HPP and was effective in prolonging life span [12, 13]. In 2015, therapy with TNALP-D₁₀ was approved for the treatment of HPP in Canada, Europe, Japan, and the USA. Although this therapy is effective in extending life and enabling patients to be weaned from artificial ventilation [14, 15], further improvements are necessary to address the following issues: (1) the HPP patient needs to receive repeated subcutaneous injections three times a week due to the short half-life of the enzyme [12], (2) the replacement enzyme dose must be increased as the patient grows, and (3) they have to continue treatment for life [16].

To resolve these problems, several approaches have been taken to establish a gene-based therapy to replace the enzyme. We have explored this approach using various viral vectors in Akp2^−/− (TNALP-knockout) HPP model mice, as they exhibit dyspnea, seizures, growth failure, and bone hypoplasia, resulting in death within 3 weeks of birth [17]. Recently, Nakano et al. showed that gene correction of induced pluripotent stem cells (iPS cells) isolated from two HPP patients rescued ALP activity and mineralization in vitro [18], suggesting a possible application of gene-corrected iPS cells for patients with HPP. These in vivo and in vitro experimental systems provide opportunities to develop a new therapy that complement the disadvantages of ERT.

Our previous studies demonstrated that a single injection of adeno-associated virus (AAV) vector, which is not pathogenic to humans and is very safe, expressing TNALP-D₁₀ in an Akp2^−/− HPP model mice successfully extended survival and corrected skeletal phenotype [19, 20]. However, several insufficient therapeutic effects were observed in skeletal tissues including hypomineralization of bone and cartilage, irregularly arranged trabeculae, partial cortical bone defects, and abnormal chondrocyte layer proliferation [20]. Although these problems are not directly life-threatening, improved treatment methods are required to prevent severely diminished physical activity.

We conjectured the non-completeness of therapeutic effects observed in the previous study [20] might be caused by insufficient local replacement of ALP in the bone and cartilage. In this study, to validate our assumption, we determined the optimal AAV vector dose for increasing the local ALP concentration in bone and mitigating femoral elongation, morphological irregularity and hypomineralization. Furthermore, we investigated the efficacy of this dose in terms of mechanical properties of the femur and spontaneous locomotor activity.

In previous studies, we successfully prolonged the survival of HPP mice by TNALP-D₁₀ replacement via the single administration of lentivirus vector [31] or an AAV vector [19, 20]. However, detailed analysis of the femurs of these mice revealed issues, including morphological irregularity, insufficient extension, and hypomineralization, suggesting that the condition was not fully resolved and that the AAV vector dosage may have been suboptimal [20]. Considering the application to patients with HPP, these results suggested that suboptimal dose would not prevent patients from suffering from growth impairment leading to short stature and persistent susceptibility to fractures. Because the most severe type of HPP affects infants and young children at an age when they are growing rapidly, growth impairment causing short stature is psychologically distressing for both patients and their families, and persistent susceptibility to fractures leads to skeletal deformity and ultimately contributes to impaired physical function. Accordingly, although these problems are not life-threatening, they severely diminish the QOL of patients and their families [32‐34]. Such impaired growth has already been observed in patients with HPP receiving the current enzyme replacement therapy, with data showing that their height and weight are greater than two standard deviations below the mean [14]. There is, therefore, a need to develop new treatment regimens capable of ameliorating the insufficient elongation, morphological irregularity, and hypomineralization of the femur by comparing the effects of ERT.

In this study, to investigate the optimum AAV vector dose needed to ameliorate femoral morphological irregularity and hypomineralization, we administered four different doses of scAAV8-CB-TNALP-D₁₀ (1.5 × 10¹¹, 7.5 × 10¹¹, 1.5 × 10¹², or 4.5 × 10¹² v.g./body) by intramuscular injection into both quadriceps femoris muscles of HPP mice. All these doses increased serum ALP activity to over 1 U/mL, thereby prolonging survival of the mice. This result is consistent with previous reports [19, 20]. We observed high vector copy numbers in liver and injected muscle, and high-ALP activity in muscle. It is known from clinical trials in hemophilia B patients that transduction does not result in long-term sustained gene expression in the liver because of the immune reaction to the viral capsid [35]. On the basis of this observation, we assume that, although transduction of AAV vector is easily achieved in muscle and liver [21], it does not persist in the liver.

Normal body weight and femoral length were achieved only in HPP mice that received a dose of 4.5 × 10¹² v.g./body; those that received lower doses exhibited plateaus in weight gain and impaired femoral elongation. In previous discussions on enzyme replacement therapy in HPP mice, it has been argued that the lack of adipose tissue in treated mice results in a plateauing of weight gain that must be due to a mechanism other than morphological irregularities of bone [36]. However, data from this study showed that, like the control mice, HPP mice treated with 4.5 × 10¹² v.g./body had normal-length femurs, and their weight gain did not plateau but continued to increase until they ultimately reached normal weight. These results suggest that persistent insufficient bone elongation is likely a direct cause of the plateau in weight gain. We performed a micro-CT analysis to observe the femoral morphology in greater detail. HPP mice treated with 1.5 × 10¹¹ v.g./body, which resulted in a serum ALP activity of 1 U/mL and was found to extend survival, exhibited morphological irregularities, including persistent epiphyseal cupping, irregularly arranged trabeculae, cortical bone defects, and non-uniform cortical bone thickness in previous studies [20]. Cancellous bone analysis confirmed that the disease also remained unresolved in quantitative terms. In contrast, HPP mice treated with 4.5 × 10¹² v.g./body exhibited bone morphology that was almost identical to that of the control mice, and the values measured in the cancellous bone analysis and FEA also improved. These findings support the hypothesis that the level of serum ALP activity required to prolong survival (1 U/mL) is insufficient in terms of local ALP replacement in femoral bone. At this serum ALP level, the disease remains unresolved in a number of aspects, including insufficient elongation of the femur, epiphyseal cupping, irregularly arranged trabeculae, hypomineralization, and cortical bone defects. It also suggests that administering TNALP-D₁₀ at a sufficiently high dose for local femoral bone may ameliorate these unresolved problems.

To evaluate the histological condition of bone, we conducted Alcian blue staining, H&E staining, and ALP staining. In HPP mice treated with 1.5 × 10¹¹ v.g./body, histological staining revealed morphological irregularities in the growth plate cartilage and articular cartilage as well as ectopic fibrous tissue in the epiphysis and cortical bone, and there were only a few weak positive spots in the ALP staining. In contrast, mice treated with 4.5 × 10¹² v.g./body exhibited the same morphology and ALP distribution as those seen in the control group. This result lends further support to the hypothesis that TNALP-D₁₀ replacement at a dose sufficient to achieve normal levels of ALP in local bone brings about normal calcareous degeneration of the hypertrophic chondrocyte layer, leading to normal endochondral ossification [37], and ameliorates insufficient elongation, morphological irregularity, and hypomineralization of the femur. In the present study, we applied the frozen sections (10 µm thick) obtained from unfixed and undecalcified bone tissues to detect more accurate ALP activity. This technique provided successful detection of ALP activity, but it failed to observe details of chondrocyte columnar arrangement due to the reduced resolution of histology. Application of more advanced technique to detect accurate ALP activity retaining detailed morphology will be necessary to compare accurate therapeutic effects between ERT and gene therapy at histological level.

In the bone strength analysis of the epiphyseal ROI in HPP mice treated with 1.5 × 10¹¹ v.g./body, we found that the growth plate and trabeculae were unable to absorb the mechanical stress generated by forced displacement and exhibited extremely fragile structural characteristics compared with those of control mice. In contrast, in HPP mice treated with 4.5 × 10¹² v.g./body, which had bone morphology almost identical to that of the control group, these areas were able to withstand displacement to a level comparable to that of the control group. This finding suggests that the persistence of cortical bone defects, reduced trabecular numbers, and irregularly arranged trabeculae observed in HPP mice treated with suboptimal doses of vector may interfere with one of the most important functions of bone—its role as a supporting tissue maintaining the body’s structure against its own weight.

Although femoral morphology was improved in HPP mice treated with 4.5 × 10¹² v.g./body, histological analysis unexpectedly revealed the presence of ectopic bone structures in these mice, which will be an adverse effect of the treatment. We speculate that these structures relate to the timing of the start of treatment. Because treatment began neonatally, endochondral ossification in utero went untreated, and hypertrophic chondrocytes did not undergo calcareous degeneration, and overproliferated. As a result, overproliferated cartilages of the epiphysis may have separated when force was applied. The ectopic calcification seen in the HPP mice treated with 4.5 × 10¹² v.g./body may have been formed by TNALP-D₁₀ replacement in this detached cartilage. This issue might be resolved by starting the treatment in utero, and we are now planning to conduct fetal gene therapy experiments. Here, we found that a high dose of 4.5 × 10¹² v.g./body was required to provide sufficient ALP replacement in local bone to ameliorate the signs of HPP. In gene therapy using large animals, intravenous administration of high doses of AAV vectors reportedly causes liver toxicity [38, 39]. This toxicity is likely caused not only by the presence of contaminating proteins or the method of viral vector preparation [40, 41] but also by the viral load or promotor sequence [39]. Previous studies have found that the treatment vector is more easily introduced into bone when gene therapy is conducted in utero rather than neonatally [42]. Investigating in utero gene therapy could enable a lower dose of the vector to be administered, which is desirable from a safety perspective.

The symptoms of HPP patients include muscle weakness and muscle pain, and many patients exhibit delayed or impaired motor development [43, 44]. Here, although there was no significant difference in activity levels on the floor between the control mice and HPP mice treated with AAV vector at 1.5 × 10¹¹ or 4.5 × 10¹² v.g./body, we found that the activity in the wheel cage was lower in HPP mice at all treatment doses examined than in control mice. We previously reported that activity was restored in the treated HPP mice [20]; however, in that study, we performed only a short-term analysis of activity levels on the floor. Our investigation of daily spontaneous activity in a wheel cage provided the important knowledge that the activity level does not, in fact, recover, even in mice treated with 4.5 × 10¹² v.g./body. Given our finding that the structure and strength of bone did recover in these mice, we considered that joint or muscle abnormalities may be present. Attempts have recently been made to create large-animal models of HPP, and the successful establishment of an ovine HPP model using CRISPR/Cas9 has been reported [45]. At 2 months of age, the HPP model sheep showed signs of muscle weakness, with a qualitatively altered gait, and skeletal muscle biopsy revealed abnormalities in muscle fiber size and incorrectly folded mitochondrial cristae [45]. Myopathy has also been reported in HPP patients, with muscle biopsy revealing no abnormalities in some cases [46] but abnormalities in muscle fiber size in others [43]. This mechanism is currently under investigation.

In this study, injection of model HPP mice with scAAV8-CB-TNALP-D₁₀ at an optimal high dose resulted in high levels of bone ALP activity and induced normal growth and improved bone formation. Therefore, optimized AAV vector-mediated TNALP-D₁₀ replacement therapy is a promising option for improvement of the QOL of patients with the infantile form of HPP. Further studies concerning the comparison of the therapeutic effect between our gene therapy and ERT are necessary using Akp2^−/− mice before clinical application.