Intra-articular injection of mesenchymal stem cells expressing coagulation factor ameliorates hemophilic arthropathy in factor VIII-deficient mice

Authors


Tsukasa Ohmori or Yoichi Sakata, Research Division of Cell and Molecular Medicine, Center for Molecular Medicine, Jichi Medical University, 3111-1 Yakushiji, Shimotsuke, Tochigi 329-0498, Japan.
Tel.: +81 285 58 7397; fax: +81 285 44 7817.
E-mail: tohmori@jichi.ac.jp; yoisaka@jichi.ac.jp

Abstract

Summary.  Background:  Transplantation of cells overexpressing a target protein represents a viable gene therapeutic approach for treating hemophilia. Here, we focused on the use of autologous mesenchymal stem cells (MSCs) expressing coagulation factor for the treatment of coagulation factor VIII (FVIII) deficiency in mice.

Methods and Results:  Analysis of luciferase gene constructs driven by different promoters revealed that the plasminogen activator inhibitor-1 (PAI-1) gene promoter coupled with the cytomegalovirus promoter enhancer region was one of the most effective promoters for producing the target protein. MSCs transduced with the simian immunodeficiency virus (SIV) vector containing the FVIII gene driven by the PAI-1 promoter expressed FVIII for several months, and this expression was maintained after multiple mesenchymal lineage differentiation. Although intravenous injection of cell supernatant derived from MSCs transduced with an SIV vector containing the FVIII gene driven by the PAI-1 promoter significantly increased plasma FVIII levels, subcutaneous implantation of the MSCs resulted in a transient and weak increase in plasma FVIII levels in FVIII-deficient mice. Interestingly, intra-articular injection of the transduced MSCs significantly ameliorated the hemarthrosis and hemophilic arthropathy induced by knee joint needle puncture in FVIII-deficient mice. The therapeutic effects of a single intra-articular injection of transduced MSCs to inhibit joint bleeding persisted for at least 8 weeks after administration.

Conclusions:  MSCs provide a promising autologous cell source for the production of coagulation factor. Intra-articular injection of MSCs expressing coagulation factor may offer an attractive treatment approach for hemophilic arthropathy.

Introduction

Hemophilia is a recessive X-linked genetic bleeding disorder involving a lack of functional coagulation factor VIII (FVIII) or FIX. Hemophilia is considered to be suitable for gene therapy, because it is caused by a single gene abnormality, and therapeutic coagulation factor levels may vary across a broad range [1]. Although the objective of gene therapy is to correct a defective gene sequence responsible for a disease phenotype, recent studies have focused on ectopic expression of the target gene by viral or non-viral gene transfer [2,3]. Most gene therapy strategies for hemophilia are now exploiting two basic approaches: direct administration of a viral or plasmid vector for in vivo gene transfer, or transplantation of cells transduced ex vivo [2,3]. Adeno-associated virus (AAV) vectors have been extensively used for the former approach, and have shown dramatic efficacy in some animal models [4]. In fact, therapeutic levels of coagulation factor have also been achieved in patients with hemophilia B by use of the AAV8 serotype in a phase I clinical trial [5].

The other gene therapy strategy for hemophilia involves transplantation of cells transduced ex vivo to ectopically express coagulation factor [3]. We and others reported that transplantation of hematopoietic stem cells transduced with lentiviral vector expressing coagulation factor corrected the phenotype of mouse models of hemophilia [6–9]. In these studies, a blood cell lineage-specific promoter enabled the expression of coagulation factors in specific lineages of blood cells, including platelets [7,9], red blood cells [6], and lymphocytes [10]. The use of blood cells to deliver coagulation factor is particularly attractive, as it avoids interference from circulating inhibitors [6,11]. Autologous hematopoietic stem cell-based gene therapy now represents an emerging therapeutic option for several immunodeficiency diseases [12]. However, the requirement for a conditioning regimen, including irradiation and/or chemotherapy, for successful transplantation means that this approach is not realistic for hemophilic patients.

Local implantation of cells expressing coagulation factor without conditioning treatments has been proposed as an alternative approach for cell-based therapy for hemophilia [13–15]. The advantage of locally implanting ex vivo transduced cells is that it avoids unexpected side effects caused by systemic influx of a viral vector. Many different types of cell have been reported to effectively express coagulation factor in cell-based therapy for hemophilia [13–16]. However, the emergence of a neutralizing antibody against coagulation factor or the loss of viability of the transplanted cells often limits their clinical applications, even if transient therapeutic expression of FVIII has been achieved [13,14]. Indeed, the transplantation of autologous fibroblasts expressing high levels of FVIII onto the omentum failed to achieve long-term expression of the coagulation factor in human clinical trials [16]. Clearly, further development of transplantation procedures is necessary before cell-based therapy can be successfully applied for hemophilic patients.

In this study, we focused on the use of mesenchymal stem cells (MSCs) as an autologous cell source to produce coagulation factor for cell-based gene therapy of a mouse model of hemophilia A. MSCs can be easily expanded in vitro, and effectively produce FVIII following lentiviral transduction. We found that the plasminogen activator inhibitor-1 (PAI-1) promoter was one of the most effective promoters for producing the target protein. Although we failed to consistently increase the plasma levels of coagulation factor after subcutaneous transplantation of the transduced cells in FVIII-deficient mice, we did confirm that the transduced MSCs elicited therapeutic effects by acting as a local hemostatic biomaterial in hemarthrosis and the resultant hemophilic arthropathy.

Materials and methods

The methods for construction of luciferase reporter plasmids, the luciferase reporter assay, the isolation of murine MSCs, the differentiation of MSCs, the subcutaneous transplantation of MSCs, histologic analysis and analysis of circulating FVIII inhibitors are described in detail in Data S1.

Mice

FVIII-deficient mice (B6;129S4-F8tm1Kaz/J) [17] were kindly provided by H. H. Kazazian Jr (University of Pennsylvania, Philadelphia, PA, USA). C57BL/6J mice were purchased from Japan SLC (Shizuoka, Japan). All animal procedures were approved by the Institutional Animal Care and Concern Committee at Jichi Medical University, and animal care was in accordance with the committee’s guidelines.

cDNA cloning, construction of lentiviral vectors, and virus production

The cDNAs for human B-domain-deleted FVIII (hBDD-FVIII) were generated as previously described [18]. The cDNAs for enhanced green fluorescent protein (EGFP), luciferase or hBDD-FVIII under the control of the indicated internal promoter were cloned into a self-inactivating simian immunodeficiency virus (SIV) lentiviral vector plasmid [19]. The SIV lentiviral vectors were generated essentially as previously described [7]. The transduction units of the lentiviral vector, transgene expression and proviral integration into the genomic DNA were measured as previously described [7,20].

Measurement of coagulation factor activity and antigen expression

Human hFVIII (hFVIII) antigen (hFVIII:Ag) was measured with an anti-hFVIII-specific ELISA kit (ASSERACHROM VIII:Ag; Diagnostica Stago, Seine, France). The functional activity of hFVIII (hFVIII:C) was measured with an activated partial thromboplastin time-based, one-stage clotting-time assay on an automated coagulation analyzer (Sysmex CA-500 analyzer; Sysmex, Kobe, Japan). We used pooled normal human plasma as a reference to measure both hFVIII:C and hFVIII:Ag.

Bioluminescence studies

The fates of transduced cells in identical recipient mice in vivo were directly assessed by measuring luciferase activities derived from the transduced cells (IVIS Imaging System and Living Image software; Xenogen, Alameda, CA, USA), as previously described [20].

Hemarthrosis model and intra-articular injection

The mouse model of hemophilic hemarthrosis was established by single needle puncture of the knee joints of FVIII-deficient mice, as previously described [21,22]. Briefly, 6–8-week-old mice were anesthetized with isoflurane, and the hair covering the left knee joint was shaved. The knee joint capsule was punctured with a 30-G needle below the patella to induce intra-articular bleeding. MSCs (1 × 105 cells per 5 μL) or vehicle alone (5 μL of saline) were directly injected into the affected knee joint with a Hamilton syringe (Hamilton, Bonaduz, Switzerland). The mice were allowed to recover. Then, at specified times after surgery, the mice were anesthetized with isoflurane, perfused with 50 mL of saline, and killed. Knee joints were collected by sectioning the femur and tibia, and macroscopic bleeding was photographed. Some knee sections were fixed and decalcified by the use of routine histologic procedures.

Quantification of hemoglobin content

Soft tissue around the knee joint was homogenized in distilled water, and processed for the measurement of tissue bleeding as previously described [23]. Briefly, 20 μL of supernatant containing hemoglobin was incubated with 80 μL of Drabkin’s reagent (Sigma Aldrich Co., St. Louis, MO, USA), and the hemoglobin concentration was assessed by measuring the OD of the solution at 550 nm.

Grading of arthropathy pathology

Hemophilic arthropathy was graded according to a verified scoring system [22]. In this system, evidence of synovial overgrowth (0–3), neovascularity (0–3), the presence of blood (0 or 1), discoloration by hemosiderin (0 or 1), synovial vilus formation (0 or 1) or cartilage erosion (0 or 1) is scored from 0 to 10. Independent reviewers blinded to the experimental conditions examined the entire joint space and articular surfaces of the sections from each knee. The area of greatest synovial thickening and vascularity was identified, and mean total synovitis scores at the region from each knee were determined. Images were captured with a charge-coupled device camera by the use of NIS-Elements software (Nikon, Tokyo, Japan).

Results

The PAI-1 promoter enables efficient transgene expression in MSCs

We first examined whether MSCs could release functional hFVIII. MSCs, mouse embryonic fibroblasts (MEFs), and HepG2 cells, a hepatocellular carcinoma cell line, were transduced with the SIV lentiviral vector expressing hBDD-FVIII driven by a cytomegalovirus (CMV) promoter. MSCs efficiently produced functional hFVIII, as compared with other cell types (Fig. S1). The relative coagulant activities (hFVIII:C/hFVIII:Ag) were 1.44 ± 0.297, 0.90 ± 0.246, and 1.420 ± 1.041 in MSCs, MEFs, and HepG2 cells, respectively (Fig. S1C).

To achieve efficient expression of the transgenes in MSCs, we compared the activities of several promoters in MSCs by lipofection. Figure 1A shows a schematic diagram of the promoters used in the experiment. The luciferase reporter gene was used to compare the promoter activities, and the luciferase activity of the promoter was normalized for the luciferase activity of the SV40 promoter with an enhancer sequence. We chose the PAI-1 promoter as a candidate promoter in MSCs, because the PAI-1 gene is a highly inducible gene in MSCs exposed to hypoxia [24]. The DNA fragments for the promoter region of the human PAI-1 gene were fused with the early enhancer element of the CMV promoter, as previously described [25]. As shown in Fig. 1B, the CMV and PAI-1 promoter enabled efficient expression of luciferase in MSCs.

Figure 1.

 Comparison of promoter activities in mesenchymal stem cells (MSCs). (A) Schematic diagrams of the promoters used in the experiments. (B) Each construct, along with a promoterless vector (basic) or a positive control vector (SV40/Enhancer), was transfected into MSCs. Luciferase activity was measured 48 h after transfection, and is shown relative to the activity driven by the SV40 promoter (SV40/Enhancer). Each experiment was carried out four to six times with duplicate samples. Values are means ± standard deviations. CE–PAI-1, PAI-1 promoter coupled with CMV promoter enhancer region; CMV, cytomegalovirus; EF1α, elongation factor-1α; PAI-1, plasminogen activator inhibitor-1; PGK, phosphoglycerate kinase 1.

Next, we constructed SIV-based lentiviral vectors containing the EGFP gene under the control of either the CMV promoter (SIV–CMVp–EGFP) or the PAI-1 promoter (SIV–PAI-1p–EGFP) to confirm that gene expression is efficiently driven by the candidate promoters in MSCs with the SIV vector. Both vectors efficiently transduced the EGFP gene into MSCs (Fig. 2). It is of note that the mean fluorescence intensity (MFI) of EGFP expression driven by the PAI-1 promoter was much greater than the MFI of expression driven by the CMV promoter (Fig. 2). EGFP expression was maintained for at least 9 weeks after transduction (Fig. 2). Therefore, we used the PAI-1 promoter in subsequent experiments because of its efficient expression of the transgene in MSCs by SIV.

Figure 2.

 Expression of enhanced green fluorescent protein (EGFP) in mesenchymal stem cells (MSCs) transduced with simian immunodeficiency virus (SIV) vectors carrying the EGFP gene driven by the cytomegalovirus (CMV) or plasminogen activator inhibitor-1 (PAI-1) promoter. MSCs were transduced with SIV–CMVp–EGFP (CMVp–EGFP) or SIV–PAI-1p–EGFP (PAI-1p–EGFP). Cellular expression of EGFP was analyzed by flow cytometry. (A) Representative data showing EGFP expression after transduction at the indicated multiplicity of infection (MOI). (B–E) The percentage of EGFP-positive cells (B, C) or the mean fluorescence intensity (MFI) of EGFP (D, E) was quantified in cells transduced with an increasing MOI for 48 h (B, D), or with a fixed MOI (30) for various times (C, E). Values are means ± standard deviations (n = 5).

hFVIII expression in MSCs during passage and differentiation

We next examined the maintenance of hFVIII production after transduction during passage. MSCs were transduced with the SIV vector containing hBDD-FVIII under the control of the PAI-1 promoter (SIV–PAI-1p–hFVIII) at an indicated multiplicity of infection (MOI). The activity of hFVIII produced from the cells over 24 h was assessed every week before each passage. As shown in Fig. 3A, the cells produced hFVIII in a vector dose-dependent manner, and the production was stably maintained in vitro for at least 9 weeks. Proviral integration into the genome was significantly correlated with hFVIII:C after transduction in a linear regression model (< 0.0001) (Fig. S2).

Figure 3.

 Persistent expression of human FVIII (hFVIII) from transduced mesenchymal stem cells (MSCs) during maintenance and differentiation. (A) MSCs were transduced with the simian immunodeficiency virus (SIV) vector carrying the hFVIII gene driven by the plasminogen activator inhibitor-1 promoter at the indicated multiplicity of infection (MOI). hFVIII activity (hFVIII:C) in the supernatant derived from 5 × 105 cells for 24 h in 1 mL was assessed at the indicated times after transduction with a one-stage clotting-time assay on an automated coagulation analyzer. Values are means ± standard deviations (SDs) (n = 4). (B) MSCs transduced with the SIV vector carrying the hFVIII gene at an MOI of 30 were differentiated into multiple mesenchymal lineages, as described in Data S1. Immunocytochemistry was performed to detect differentiation into adipocytes (anti-FABP-4), osteocytes (anti-osteopontin), or chondrocytes (anti-collagen II) (red), and hFVIII antigen (hFVIII:Ag) (anti-hFVIII polyclonal antibody; green). Nuclear localization was simultaneously examined by 4′,6-diamidino-2-phenylindole (DAPI) staining. The merged images show colocalization of lineage marker and hFVIII antigen. Scale bars: 60 μm. (C–E) The supernatants were isolated at the indicated times after adipogenic (C), osteogenic (D) or chondrogenic differentiation (E). The antigen levels of hFVIII in the supernatant were quantified by ELISA. Values are means ± SDs (n = 4). FABP-4, fatty acid binding protein-4.

To investigate whether differentiation of MSCs affects hFVIII production, MSCs transduced with SIV–PAI-1p–hFVIII at an MOI of 30 were differentiated into adipocytes, osteocytes and chondrocytes in vitro. We confirmed the expression of each lineage-specific marker and hFVIII:Ag after differentiation (Fig. 3B). Under the same conditions, the production and secretion of hFVIII:Ag persisted during adipogenic and osteogenic differentiation (Fig. 3C,D). On the other hand, although hFVIII:Ag was secreted from the cells during chondrogenic differentiation, the level was somewhat lower (Fig. 3E). This was probably because of the culture conditions, as the cells were cultured as aggregate cell pellets in chondrogenic differentiation medium (see Data S1). These results suggest that the release of hFVIII is maintained during cell division in undifferentiated MSCs, and that lineage differentiations are unlikely to influence the production of hFVIII.

Subcutaneous transplantation of MSCs expressing hFVIII does not improve the phenotype of FVIII-deficient mice

We next investigated the therapeutic effects of transplanting engineered MSCs expressing hFVIII on systemic bleeding in FVIII-deficient mice with hemophilia A. Direct intravenous injection of concentrated supernatant from 0.4–4 × 106 MSCs transduced with SIV–PAI-1p–hFVIII significantly increased plasma hFVIII:Ag levels (Fig. S3A), indicating that the autologous MSCs could be an attractive cell source for the production of functional FVIII. However, subcutaneous implantation of transduced MSCs mixed with Matrigel resulted in a marginal increase in plasma FVIII levels, and the expression of hFVIII:Ag was not persistent, even if the number of cells was increased to 3 × 107 (Fig. S3B). As expected from previous reports [14], we detected the emergence of circulating plasma inhibitors of hFVIII after transplantation (Fig. S3C). Thus, we concluded from these results that subcutaneous transplantation of transduced MSCs was unable to significantly improve systemic bleeding in mice with hemophilia A.

Intra-articular injection of MSCs expressing hFVIII ameliorates hemarthrosis and arthropathy in FVIII-deficient mice

We next examined whether transduced autologous MSCs could serve as a local hemostatic biomaterial. The most significant morbidity resulting from congenital FVIII deficiency is the progressive destruction of joints resulting from recurrent intra-articular hemorrhage. The coagulation factor derived from transduced MSCs may prevent local hemorrhage, and the ability of MSCs to differentiate into osteocytes and chondrocytes might promote repair of the affected joints. Therefore, we examined the effects of intra-articular injection of transduced MSCs in preventing intra-articular hemorrhage and hemophilic arthropathy.

We first examined the biodistribution of transduced MSCs after injection into the knee. The MSCs were efficiently transduced with the SIV vector having a luciferase gene under the control of the PAI-1 promoter (Fig. 4B,C). We next injected 100 000 transduced MSCs into the left knee articular space in C57BL/6J mice (0.33% of the cells used in subcutaneous transplantation). The intensity and biodistribution of luciferase expression derived from the cells were imaged at the indicated times after injection. As shown in Fig. 4D,E, luciferase derived from the transduced MSCs was detected in the injected knee, and was maintained for at least 4 weeks. In contrast, no luciferase activity was detected in other organs (Fig. 4D). Furthermore, real-time RT-PCR could not detect transgene expression in other organs, including the heart, lung, liver, kidney, spleen, and bone marrow, after transplantation (data not shown). Immunohistochemical staining for luciferase in the knee revealed that luciferase was mainly expressed in chondrocytes in the joint structure (Fig. S4), suggesting that the transduced MSCs differentiated into chondrocytes after injection.

Figure 4.

 Fate of transduced mesenchymal stem cells (MSCs) injected into the knee joint space in vivo. (A) Schematic diagram of the simian immunodeficiency virus (SIV) lentiviral vector used in this experiment. The SIV vector expresses a luciferase gene driven by the plasminogen activator inhibitor-1 (PAI-1) promoter. (B) MSCs were transduced with the SIV lentiviral vector at a multiplicity of infection (MOI) of 30. Ex vivo bioluminescence images of transduced MSCs were obtained with the IVIS Imaging System. (C) MSCs were transduced with the SIV lentiviral vector at the indicated MOI, and ex vivo bioluminescence of the transduced cells was quantified (photons s–1). Values are means ± standard deviations (SDs) (n = 3). (D) In vivo bioluminescence images of transduced MSCs after transplantation. Photons transmitted through the body were recorded with the IVIS Imaging System 1 day after injection of the transduced MSCs (1 × 105 cells) into the left knee joint space. (E) In vivo bioluminescence of the mice was quantified for the indicated times after injection (photons s–1). Values are means ± SDs (n = 5). CMV, cytomegalovirus; cPPT, central polypurine tract; PAI-1p, plasminogen activator inhibitor-1 promoter; RRE, rev response element.

To examine the possibility that locally administered MSCs expressing hFVIII could protect against hemarthrosis in the absence of circulating FVIII, we injected the MSCs transduced with SIV–PAI-1p–hFVIII into the knee space (1 × 105 cells) after single needle puncture in FVIII-deficient mice. Macroscopic bleeding around the affected knee was observed, and the blood leakage was quantified as the amount of hemoglobin measured 24 h after the knee challenge. Knee joint needle puncture resulted in massive bleeding in the joint space and in peripheral tissues (Fig. 5A,C). Intravenous injection of recombinant hFVIII significantly and dose-dependently improved joint bleeding (Fig. 5A–C). Interestingly, injection of MSCs transduced with SIV–PAI-1p–hFVIII, but not of non-transduced MSCs, had hemostatic effects that were equivalent to those in mice with hemophilia A intravenously treated with 1 U per mouse of recombinant hFVIII (Fig. 5A,C). The plasma concentration of recombinant hFVIII after intravenous injection of 1 U per mouse was 20–30% of that in normal human pooled plasma (Fig. 5B). On the other hand, we could not detect hFVIII:Ag in plasma after intra-articular injection of MSCs transduced with SIV-PAI-1p-hFVIII (data not shown). The estimated hFVIII:C produced by transplanted MSCs (1 × 105 cells) was 0.025–0.05 U per 24 h. We also found that intra-articular injection of 0.1 U of recombinant hFVIII, but not 0.01 U, significantly inhibited hemarthrosis (Fig. 5C). The peak concentrations seemed to be higher after direct injection of recombinant hFVIII than those produced by the transduced MSCs, suggesting that both hFVIII and MSCs in the synovial space are essential for the therapeutic effects of our procedure.

Figure 5.

 Local injection of transduced mesenchymal stem cells (MSCs) expressing human FVIII (hFVIII) protects against hemarthrosis induced by joint puncture in FVIII-deficient mice. FVIII-deficient mice received an intravenous dose of recombinant hFVIII or an intra-articular injection of non-transduced MSCs or transduced MSCs expressing hFVIII (1 × 105 cells), and this was followed by joint capsular needle puncture injury. (A) Macroscopic findings of hemarthrosis at 24 h after knee puncture. IV: intravenous injection of the indicated dose of recombinant hFVIII. IA: intra-articular administration of MSCs transduced without (MSCs) or with (hFVIII MSCs) SIV–PAI-1p–hFVIII. (B) Plasma hFVIII antigen (hFVIII:Ag) levels at the indicated times after intravenous administration of recombinant hFVIII. Values are means ± standard deviations (SDs) (n = 3). (C) Bleeding around the knee joint quantified as the hemoglobin (Hb) concentration. Values are means ± SDs (n = 5). Data are also shown for mice receiving an intra-articular injection of recombinant hFVIII (0.01 U or 0.1 U) (n = 5). **< 0.01 as compared with the untreated control (two-tailed Student’s t-test). PAI-1p, plasminogen activator inhibitor-1 promoter; SIV, simian immunodeficiency virus vector.

We next assessed the progression of hemophilic arthropathy after intra-articular injection of transduced MSCs. Four weeks after injection, the joints were harvested, and histopathologic grading of arthropathy was performed. Results were compared between mice treated with or without intravenous injection of recombinant hFVIII or with intra-articular injection of non-transduced MSCs. As shown in Fig. 6A, intra-articular injection of MSCs transduced with SIV–PAI-1p–hFVIII significantly reduced the extent of hemorrhage-induced synovitis, including synovial hyperplasia, vascularity, and discoloration. The pathologic score of mice given an intra-articular injection of transduced MSCs was equivalent to that in mice intravenously treated with 1–4 U per mouse of recombinant hFVIII (Fig. 6B).

Figure 6.

 Local injection of transduced mesenchymal stem cells (MSCs) expressing human FVIII (hFVIII) protects against hemophilic arthropathy in FVIII-deficient mice. (A) Representative histopathologic images taken 4 weeks after the joint challenge. Without injury: FVIII-deficient mouse knee joint without knee puncture. With IA saline: the knee joint after knee puncture and treatment with intra-articular saline. With IA hFVIII MSCs: the knee joint after knee puncture and treatment with intra-articular MSCs transduced with SIV–PAI-1p–hFVIII. Higher magnifications of the numbered boxed regions are shown in the lower panel. Scale bars: 125 μm in upper panel; 50 μm in lower panel. (B) Histologic changes were assessed with a validated mouse hemophilic synovitis grading system. The severity of synovial hyperplasia, vascularity, or the presence of blood, synovial villus formation, discoloration by hemosiderin or cartilage erosion are graded from 0 to 10. Control: FVIII-deficient mouse knee joint without knee puncture. IV: the knee joint after knee puncture and treatment with intravenous injection of the indicated dose of recombinant hFVIII. IA: knee joint after knee puncture with intra-articular injection of MSCs transduced without (MSCs) or with (hFVIII MSCs) SIV–PAI-1p–hFVIII. Values are means ± standard deviations (SDs) (n = 4). **< 0.01 as compared with untreated control (two-tailed Student’s t-test).

We also investigated the long-term treatment effects of a single intra-articular injection of transduced MSCs in inhibiting joint bleeding. FVIII-deficient mice received an intra-articular injection of transduced MSCs expressing hFVIII, followed by rechallenge (i.e. needle puncture of the affected knee). As shown in Fig. 7B, hemarthrosis at 24 h after the rechallenge was significantly improved by intra-articular injection of transduced MSCs. The therapeutic effects persisted for at least 8 weeks after administration (Fig. 7A). Low titers of circulating inhibitors of hFVIII could be detected after intra-articular injection of transduced MSCs, but these were much lower than those detected after subcutaneous transplantation of MSCs (Figs 7B and S3C).

Figure 7.

 Intra-articular (IA) injection of transduced mesenchymal stem cells (MSCs) expressing human FVIII (hFVIII) persistently inhibits hemarthrosis, and protects against hemarthrosis in the presence of low titers of neutralizing antibodies against hFVIII. (A) FVIII-deficient mice received an IA injection of transduced MSCs expressing hFVIII (1 × 105 cells), and this was followed by joint capsular needle puncture at the indicated times after IA injection. A schematic diagram of the procedure is shown in the box. Hemarthrosis at 24 h after knee puncture was quantified as the hemoglobin (Hb) concentration. Values are means ± standard deviations (SDs) (n = 4–9). *< 0.05 as compared with the saline-injected control group (two-tailed Student’s t-test). (B) Circulating plasma inhibitors were assessed as Bethesda Units (BU) mL–1 at the indicated times after IA injection. Values are means ± SDs (n = 4–10). (C) Neutralizing antibodies against hFVIII (2, 10 or 50 BU per mouse) were intravenously administrated into FVIII-deficient mice. Circulating plasma inhibitor concentrations were assessed as BU mL–1 (n = 6). (D) FVIII-deficient mice received an IA injection of transduced MSCs expressing hFVIII (1 × 105 cells or 1 × 106 cells), and this was followed by joint capsular needle puncture. Hemarthrosis at 24 h after knee puncture was quantified as the Hb concentration. A schematic diagram of the procedure is shown in the box. Values are means ± SDs (n = 3). *< 0.05 as compared with the saline-injected control group (two-tailed Student’s t-test).

We finally examined whether intra-articular injection of transduced MSCs ameliorates hemarthrosis in the presence of circulating inhibitors. FVIII-deficient mice were immunized by weekly injection of recombinant hFVIII (4 U per mouse). We obtained pooled plasma containing a high titer of hFVIII inhibitor after six doses (1110 Bethesda Units [BU] mL–1), and intravenously injected the indicated volume of plasma into naïve FVIII-deficient mice. The plasma neutralizing antibody titer increased to 1.62 ± 0.387, 7.58 ± 0.577 and 35.14 ± 23.460 after the injection of 2, 10 and 50 BU per mouse, respectively (Fig. 7C). Intra-articular injection of transduced MSCs (1 × 105 cells) reduced the hemarthrosis elicited by needle puncture in the presence of a low titer of the inhibitors (2 BU per mouse), although the effects were weaker in the presence of a higher titer of circulating inhibitor (10 or 50 BU per mouse) (Fig. 7D). Increasing the number of transplanted cells (1 × 106 cells) partly overcame the attenuated treatment effects caused by a higher neutralizing antibody titer (Fig. 7D).

Discussion

Hemophilic arthropathy – the progressive destruction of the joint structure resulting from recurrent intra-articular hemorrhage – is a frequent and serious complication experienced by patients with severe hemophilia [26]. Despite advances in treatment and the delivery of comprehensive care, joint bleeding and hemophilic arthropathy are still the most common complications of hemophilia, and are associated with a very poor quality of life [27]. It was shown that episodic prophylactic treatment with recombinant coagulation factor could prevent joint damage in young children with severe hemophilia [28], although this approach did not prevent the progression of joint damage in adolescence, after the joint damage had fully developed [29]. The costs to the healthcare system of treating hemophilia are substantial, because of the need for prophylactic treatment with recombinant coagulation factor. Patients also experience significant loss of productivity and greatly diminished quality of life, as a result of bleeding into the joints and arthropathy [30]. Therefore, there is a need for new adjunctive treatments or prophylactic strategies that are specific for joint bleeding and the prevention of hemophilic arthropathy.

Here, we found that intra-articular transplantation of autologous MSCs expressing hFVIII ameliorated acute joint bleeding and the resultant hemophilic arthropathy in FVIII-deficient mice with hemophilia A. Intra-articular injection of transduced MSCs effectively inhibited acute joint bleeding, even in conditions where the plasma FVIII levels did not increase. It was also reported that direct injection of an AAV vector expressing FIX into the joint space improved hemophilic arthropathy in FIX-deficient mice [22]. As compared with intravenous administration of recombinant hFVIII, the main advantage of cell-based therapy and gene therapy is consistent production of the functional coagulation factor by the transduced cells. The major mechanism by which transduced MSCs affect hemophilic arthropathy seems to involve hemostasis by targeting of acute bleeding through extracellular production of hFVIII. There are also several reasons why MSCs should be selected to treat hemophilic arthropathy. MSCs can be expanded in vitro as autologous cells, can differentiate into chondrocytes and osteoblasts, and can produce a number of bioactive mediators with regenerative effects. These functions of MSCs can be exploited therapeutically to repair degenerative joints, as MSC-based strategies can be used to repair chondral and osteochondral lesions, or to modulate endogenous factors that enhance regenerative processes in degenerative joints [31]. In addition, it has been reported that MSCs can modulate immune responses and control inflammation by targeting T lymphocytes [32]. Inflammatory responses, including cytokine release and inflammatory cell invasion, caused by the response to blood in the joint space play key roles in the pathophysiology of hemophilic arthropathy [33]. The injection of MSCs into the joint space is likely to ameliorate the inflammatory response, which would otherwise promote destruction of the joint structure [31,34].

MSCs offer a promising autologous cell source for the production of coagulation factor. We found that the PAI-1 promoter was one of the most effective promoters for producing the target protein by lentiviral transduction. The different results obtained with transient transduction with a plasmid vector and lentiviral transduction may be attributable to post-transcriptional silencing of the CMV promoter after lentiviral transduction. It is well known that CMV promoter silencing limits its usefulness in many research applications and in gene therapy [35,36]. The PAI-1 promoter stably and effectively drove transgene expression even after multiple mesenchymal lineage differentiation in vitro, and luciferase expression from the transduced cells was detected at least 4 weeks after injection of the cells into the joint space. PAI-1 was reported to be an inducible factor whose expression is consistently upregulated by ischemic conditions in MSCs [24]. As hypoxia is an important event in the perpetuation of joint destruction [37], it is possible that the increase in transgene expression driven by the PAI-1 promoter under hypoxic conditions further enhances coagulation factor expression to ameliorate hemarthrosis.

Several studies have focused on cell-based therapy with MSCs expressing coagulation factor to treat hemophilia by increasing the plasma levels of coagulation factor [14,38–40]. However, long-term protein production from MSCs was not achieved in vivo after transplantation, because of the loss of cell viability and/or the emergence of inhibitory antibodies [14]. Recently, Coutu et al. [39] successfully achieved long-term expression of FIX by implanting a three-dimensional porous scaffold containing gene-modified MSCs to increase graft survival. They used the murine R333Q model of hemophilia B, which avoids the development of inhibitory antibodies [39]. In addition, Porada et al. [40] described the interesting treatment effect of MSCs in a sheep model of severe hemophilia A. They intraperitoneally transplanted MSCs expressing porcine FVIII into sheep with hemophilia A [40]. An increase in plasma FVIII activity could not be detected, and titers of the inhibitory antibody against hFVIII and porcine FVIII dramatically increased [40]. Nevertheless, transplantation of MSCs expressing coagulation factor resolved hemarthrosis and improved joint function [40]. The authors also observed the migration of transduced MSCs into a number of organs, including the synovium [40]. In our study, the level of circulating inhibitors of hFVIII induced by intra-articular injection of transduced MSCs was much lower than that following subcutaneous transplantation of MSCs, although a low titer of BUs was observed. Our results also suggest that implanting engineered MSCs expressing coagulation factor into the synovial joint space ameliorates hemarthrosis, even in the presence of inhibitory antibodies. Furthermore, a small number of transduced cells might be sufficient to achieve therapeutic effects, as compared with systemic transplantation of transduced cells. Accordingly, we believe that intra-articular injection of transduced MSCs represents a more realistic approach to ameliorate hemarthrosis and arthropathy, because of several advantages, including minimally invasive surgical procedures, the need for a small number of transduced cells, and a lower titer of inhibitory antibodies following treatment.

One of the main barriers to implementing clinical trials of gene and cell-based therapy is concern over the safety of viral vectors. We used the third generation of the SIV lentiviral vector to express coagulation factor in MSCs, because it has a better safety profile than gamma retroviral vectors (γRVs) [41]. As compared with γRVs, lentiviral vectors preferentially integrate within active transcription units without an obvious bias for proliferation-associated genes or transcriptional start sites, suggesting that lentiviral vectors are less likely to trigger oncogenic events [42]. Self-inactivating vector systems, in which the promoter activity in the U3 region of the viral long-terminal repeat (LTR) is deleted, have been used in many studies because the promoter activity of the viral LTR is associated with transcriptional activation of oncogenes in γRVs [43,44]. It is possible that the use of a physiologic promoter, such as the PAI-1 promoter in a self-inactivating vector, may be safer than using a ubiquitous viral promoter. We did not observe any tumorigenesis in the transplanted sites or abnormal proliferation of the transduced MSCs during the observation period. We believe that the safety of cell-based therapies could be further enhanced by several approaches. First, we can investigate the proviral integration sites of the transduced cells before using cell-based therapy, but not after direct injection of a viral vector. Second, we can improve the safety of cell-based therapy by blocking the cell cycle of transduced MSCs before transplantation by irradiation or pretreatment with a cytotoxic agent such as mitomycin C, if repeated injections of transduced MSCs are possible.

Some limitations of this study merit discussion before the clinical application of this procedure. The main limitation of our work is the relatively modest improvement in prevention of hemarthrosis. Although the local concentration of hFVIII achieved by intra-articular injection of the transduced cells should be higher than that reaching the joint following intravenous infusion, needle puncture-induced hemarthrosis was not completely abolished (Fig. 5). Second, our procedure induced a low neutralizing antibody titer, suggesting the possibility that our procedure would enhance the immune responses to hFVIII in patients expressing the inhibitor, particularly those with high responder inhibitor levels. Although intra-articular injection might be effective in the presence of low circulating inhibitor titers, our procedure may be more appropriate for adults who have already undergone replacement therapy several times, and might not develop inhibitory antibodies after intra-articular injection of the transduced MSCs. Furthermore, we could not fully assess the duration of transgene expression required to inhibit hemarthrosis or the fate of the transplanted MSCs. As we could recover very little RNA from around the knee joint from mice, we could not detect transgene mRNA in the joint space (data not shown). Although we believe that the therapeutic range of FVIII expression would be maintained for at least for 8 weeks, on the basis of the results shown in Fig. 7, it is important to confirm the long-term therapeutic effect and safety of this procedure. In addition, it is important to assess transgene expression and cell fate in larger animals to determine how frequently this procedure should be conducted.

In conclusion, we have proposed a new treatment strategy for hemophilic arthropathy in which MSCs expressing coagulation factor are directly injected into the target tissue. Considering that intra-articular injection is a minimally invasive procedure and that the MSCs can facilitate repair of the damaged joint structure, the procedure described here may become an attractive approach to prevent and/or treat blood-induced joint disease in hemophilic patients. Further evaluations of cell-based therapy in larger animals (e.g. cynomolgus monkey) and of the long-term safety of lentivirally transduced cells after transplantation are necessary before these procedures can be tested in clinical trials.

Addendum

Y. Kashiwakura and T. Ohmori: designed and performed the experiments, analyzed the data, and wrote the manuscript; J. Mimuro: performed experiments, analyzed the data, and revised the manuscript; A. Yasumoto, A. Sakata, and A. Ishiwata: performed experiments; M. Inoue and M. Hasegawa: provided vital reagents and critically reviewed the manuscript; S. Madoiwa, K. Ozawa, and Y. Sakata: analyzed data and revised the manuscript.

Acknowledgements

We would like to thank K. Ohashi and K. Tatsumi (Tokyo Women’s Medical University) for very helpful discussions. We also thank N. Matsumoto and M. Ito (Jichi Medical University) for their excellent technical assistance. Part of this manuscript was presented at the XXIIIth Congress of the International Society on Thrombosis and Haemostasis, Kyoto, Japan, on 25 June 2011.

Disclosure of Conflict of Interests

This study was supported by a grant from the Japan Baxter Hemophilia Scientific Research & Education Fund; Grants-in-Aid for Scientific Research (23591427, 21591249, and 23591426); the Support Program for Strategic Research Infrastructure from the Japanese Ministry of Education and Science; and Health, Labour and Science Research Grants for Research on HIV/AIDS and Research on Intractable Diseases from the Japanese Ministry of Health, Labour and Welfare.

Ancillary