original article
© The American Society of Gene Therapy
Correction of ADAMTS13 Deficiency by In Utero Gene Transfer of Lentiviral Vector encoding ADAMTS13 Genes Masami Niiya1,*, Masayuki Endo2, Dezhi Shang1, Philip W Zoltick2, Nidal E Muvarak2, Wenjing Cao1, Sheng-Yu Jin1, Christopher G Skipwith1, David G Motto3, Alan W Flake2 and X Long Zheng1 Department of Pathology and Laboratory Medicine, The Children’s Hospital of Philadelphia and The University of Pennsylvania, Philadelphia, Pennsylvania, USA; 2Department of Surgery, The Children’s Hospital of Philadelphia and The University of Pennsylvania, Philadelphia, Pennsylvania, USA; 3Department of Internal Medicine, University of Iowa, Iowa City, Iowa, USA 1
Deficiency of A Disintegrin And Metalloprotease with ThromboSpondin (ADAMTS13) results in thrombotic thrombocytopenic purpura (TTP). Plasma infusion or exchange is the only effective treatment to date. We show in this study that an administration of a selfinactivating lentiviral vector encoding human full-length ADAMTS13 and a variant truncated after the spacer domain (MDTCS) in mice by in utero injection at embryonic days 8 and 14 resulted in detectable plasma proteolytic activity (~5–70%), which persisted for the length of the study (up to 24 weeks). Intravascular injection via a vitelline vein at E14 was associated with significantly lower rate of fetal loss than intra-amniotic injection, suggesting that the administration of vector at E14 may be a preferred gestational age for vector delivery. The mice expressing ADAMTS13 and MDTCS exhibited reduced sizes of von Willebrand factor (vWF) compared to the Adamts13−/− mice expressing enhanced green fluorescent protein (eGFP). Moreover, the mice expressing both ADAMTS13 and MDTCS showed a significant prolongation of ferric chloride–induced carotid arterial occlusion time as compared to the Adamts13−/− expressing eGFP. The data demonstrate the successful correction of the prothrombotic phenotypes in Adamts13−/− mice by a single in utero injection of lentiviral vectors encoding human ADAMTS13 genes, providing the basis for developing a gene therapy for hereditary TTP in humans. Received 30 May 2008; accepted 17 September 2008; published online 28 October 2008. doi:10.1038/mt.2008.223
INTRODUCTION
ADAMTS13, a member of A Disintegrin And Metalloprotease with ThromboSpondin (ADAMTS) type 1 repeats family,1,2 controls the sizes of von Willebrand factor (vWF) by cleaving vWF at the
Tyr1605-Met1606 bond in the central A2 domain.3 ADAMTS13 protein consists of a metalloprotease domain, a disintegrin domain, first thrombospondin type 1 repeat, a Cys-rich domain and a spacer domain in addition to seven thrombospondin type 1 repeats and two CUB domains.2 The amino-terminal half of ADAMTS13 protease (up to the spacer domain) is found to be sufficient to recognize and cleave vWF under static and denatured conditions4 or to cleave peptide substrates such as vWF73 derived from the central A2 domain of vWF.5 The carboxyl-terminal half of ADAMTS13, however, may be required for collaborative binding and cleavage of vWF under fluid shear stress in vitro.6,7 Deficiencies of plasma ADAMTS13 activity owing to hereditary mutations of ADAMTS13 gene8 or acquired autoantibodies against ADAMTS13 protease9 results in an accumulation of “unusually large” vWF multimers,10 leading to a potentially fatal syndrome, thrombotic thrombocytopenic purpura (TTP). Profound thrombocytopenia, microangiopathic hemolytic anemia, and organ ischemia are characteristic features of TTP syndrome.11 If left untreated, the mortality rate is as high as 80–100%.11,12 Plasma infusion and/or plasma exchange is the only effective treatment available to date. Besides developing a TTP-like syndrome,13 mice lacking Adamts13 gene are prothrombotic,14,15 characterized by increased sizes of plasma vWF multimers and enhanced platelet adherence to activated endothelial cells in vivo. Intravenous infusion of recombinant human ADAMTS13 into Adamts13−/− mice reverses the prothrombotic phenotypes and protects mice against ferric chloride–induced arterial and venous thromboses.15 However, the short half-life of infused human ADAMTS13 into humans (t1/2, 2–4 days)16 or mice (t1/2, 15–20 minutes)15 makes the intravenous infusion of recombinant ADAMTS13 a less desirable strategy for a lifelong treatment of hereditary TTP. Gene therapy may offer an attractive alternative strategy to recombinant ADAMTS13 or plasma infusion for prevention and treatment of TTP, because only ~5–10% of ADAMTS13 activity (~0.05–0.1 mg/l protein) is required to induce and maintain
The first two authors contributed equally to this study. *Current address: Department of Hematology, Oncology and Respiratory Medicine, Okayama University Medical School, 2-5-1 Shikata-cho, Okayama, Okayama 700-8558, Japan Correspondence: X. Long Zheng, Department of Pathology and Laboratory Medicine, The Children’s Hospital of Philadelphia and The University of Pennsylvania Medical Center, 34th street and Civic Center Boulevard, 816G Abramson Research Center, Philadelphia, Pennsylvania 19104, USA. E-mail:
[email protected]
34
www.moleculartherapy.org vol. 17 no. 1, 34–41 jan. 2009
© The American Society of Gene Therapy
remission of prothrombotic states in humans.16 On the basis of our previous studies,17,18 we hypothesized that early gestational gene transfer using self-inactivated lentiviral vectors would achieve this therapeutic level in plasma. The prenatal approach is an appealing therapeutic option for an inherited genetic disorder that has an early onset and is potentially fatal, because it may pre-empt manifestations of the disease. In addition, early gestational delivery of the gene of interest results in transduction of stem cell compartments that are less accessible for gene transfer after birth.17,18 With the use of an integrating vector, subsequent expansion of transduced stem and progenitor cells should amplify expression of the protein of interest and result in a lifelong correction of the disease. Antigen processing by the naive fetal immune system would be expected to result in specific tolerance to the therapeutic proteins. Prenatal gene transfer has been successfully applied in a mouse model of hemophilia B.19 In this study, we report the achievement of sustained therapeutic levels of human full-length ADAMTS13 and the carboxylterminal truncated variant (MDTCS) in Adamts13−/− mice by intra-amniotic and intravascular (via vitelline vein) gene transfer. The expressed ADAMTS13 and MDTCS proteases were both biologically active in reducing vWF multimer sizes and protecting against ferric chloride–induced arterial thromboses in the mouse model. Our data support the feasibility of developing a gene therapy–based approach for hereditary TTP in humans.
RESULTS Survival rates of mice after early gestational gene transfer
The survival rate of newborns in Adamts13−/− background without surgical and in utero manipulation was estimated to be 95%. The survival rates after intra-amniotic injection (E8) of lentiviral vectors encoding enhanced green fluorescent protein (eGFP), ADAMTS13, and MDTCS into Adamts13−/− mice were ~59% (10/17), 63% (15/24), and 35% (10/29), respectively, suggesting various degrees of mortalities associated with the procedures. The survival rates after intravascular injection at E14 of vectors encoding eGFP, ADAMTS13, and MDTCS were ~79% (11/14), 55% (18/33), and 84% (16/19), respectively. The overall survival rate in mice injected at E14 was 76%, significantly higher than that in those injected at E8 (51%) (F value = 18.0, P = 0.002), suggesting that an administration of vectors at the later gestational stage (corresponding to early second trimester of human fetal development) may significantly reduce fetal losses. No gross structural abnormality was observed in any of the surviving newborns and adult mice, suggesting that administrations of lentiviral vectors encoding eGFP, ADAMTS13, and MDTCS either at E8 or E14 do not cause abnormality in prenatal and postnatal development. No spontaneous bleeding was observed in postnatal mice that received either the ADAMTS13 or MDTCS vectors.
Localization of transgene expression in mouse organs To determine sites of gene expression in mice after intra-amniotic administration of lentiviral vectors, mice were analyzed by stereoscopic fluorescent, immunofluorescent, and immunoperoxidase techniques described in more details previously.18 When vectors were administered at E8, the kidneys and skin were the predominant organs expressing transgenes, which showed bright Molecular Therapy vol. 17 no. 1 jan. 2009
Potential Gene Therapy for Hereditary TTP
and punctuated green fluorescent signals of eGFP reporter under stereoscopic fluorescent microscope (Figure 1a). Scattered fluorescent signals were also detectable in the brain but not in the liver (Figure 1a). Immunofluorescent staining with rabbit anti-GFP IgG further demonstrated that the transgene was primarily localized to epidermis and bulge cells in the hair follicles of the skin (Figure 1a(E)), similar to what we have reported previously.17,18 In kidney, the transgene was predominantly detected in the tubular structures as well as in the glomeruli (Figure 1a(F)). No GFP signal was observed in the control (Adamts13−/−) mice that did not receive lentiviral vectors (data not shown). In contrast to the localization of the E8 intra-amniotic studies, an intravenous administration of vector at E14 resulted in primarily liver transduction (Figure 1b) with minor gene delivery and expression evident as focal positive signals in the brain, heart, and kidney by stereofluorescent microscopy (Figure 1b). Immunoperoxidase staining further localized the transgene product to the muscle of the heart (Figure 1b(E)), and the hepatocytes in the liver (Figure 1b(F)). These data suggest that the tissue distribution of the transgene is highly dependent of the time when vector is administered.
ADAMTS13 and variant in mouse plasma To determine whether ADAMTS13 transduced in various tissues after intra-amniotic injection were secreted into blood circulation, the levels of ADAMTS13 protein were determined by an immunoprecipitation plus western blot. As shown in Figure 2, both ADAMTS13 and MDTCS were clearly detectable in mouse plasma at 6 weeks or longer after intra-amniotic administration of lentiviral vectors encoding ADAMTS13 and MDTCS, respectively. The similar protein bands were absent in the plasma of Adamts13−/− mice expressing eGFP alone, but present in normal human plasma and conditioned medium containing recombinant MDTCS after immunoprecipitation (Figure 2), indicating the secretion and delivery of the transduced human full-length ADAMTS13 and MDTCS into the murine plasma. Proteolytic activity toward the fluorescent peptide, FRETSvWF73, was determined serially over several months in the plasmas from Adamts13−/− mice and the mice injected in utero with vectors encoding ADAMTS13 and MDTCS at E8 and E14. As demonstrated in Figure 3, plasma from the mice receiving vectors at E8 and E14 encoding ADAMTS13 and MDTCS were able to cleave FRETS-vWF73 and GST-vWF73-H specifically at the Tyr1605Met1606 bond (data not shown). After E8 injection, ~5–7% and 60–70% of protease activity was detected in the plasma of these mice expressing ADAMTS13 and MDTCS, respectively (Figure 3a). The plasma levels of proteolytic activity in mice receiving vectors at E14 encoding ADAMTS13 (5–7%) (Figure 3b) were comparable with those receiving same vectors at E8. However, the plasma activity in E14 injected mice with the vector encoding MDTCS was approximately four- to sixfold lower than that in E8-injected mice (Figure 3b). This difference in the plasma activity may reflect the relatively lower vector dose administered into mice at E14. No proteolytic cleavage FRETS-vWF73 (Figure 3) and GST-vWF73 (data not shown) was detected in Adamts13−/− mice expressing eGFP alone, suggesting that these assays are highly specific for detecting ADAMTS13 activity in mouse plasma. The elevated proteolytic activity in plasma persisted for 12–24 weeks (the lengths of this 35
© The American Society of Gene Therapy
Potential Gene Therapy for Hereditary TTP
a
A
B
C
E
b
A
B
D
C
D
F
E
F
50 micron
50 micron
50 micron
50 micron
Figure 1 Localization of transgene products in mice receiving (a) intra-amniotic and (b) intravascular administration of lentiviral vectors. (a) A–D are the stereomicroscopic images of the fluorescent signals at birth in the (A) brain, (B) skin, (C) liver, and (D) kidney, respectively, of a representative mouse receiving vector encoding eGFP at E8; (E) and (F) are the immunofluorescent images in the skin and kidney expressing eGFP at 6 months of post-intra-amniotic injection (E8) of vector. (b) A–D are the stereomicroscopic images of the fluorescent signals (arrowheads) at birth in the (A) brain, (B) heart, (C) liver, and (D) kidney, respectively, in a mice receiving vector encoding eGFP at E14; E and F are the images (×20) of positive immunoperoxidase staining (arrowheads) of the heart and liver sections, respectively, in a mouse injected with vector encoding eGFP.
a
3
− −/
3
1 AD
1 AD
FL
P NH
D rA
b
13
− −/
13 AD
CS
DT
M
CS
CM
DT
rM
97K
195K 1 2 3 IP: anti-A13 IgG
4
1 2 3 IP: anti-A13 IgG
4
Figure 2 Recombinant full-length ADAMTS13 and MDTCS in mouse plasma. Immunoprecipitation and western blot detected full-length (a) ADAMTS13 (195K) (a, lane 2) and (b) MDTCS (b, lane 2) at 5 months of age in mice receiving intra-amniotic injection of vectors. No detectable band was seen in plasma from Adamts13 null mice (a, lane 1 and b, lane 1). Other control includes normal human plasma (NHP, a, lane 3) and the conditioned medium containing recombinant human MDTCS (b, lane 3). The recombinant full-length ADAMTS13 (a, lane 4) and MDTCS (b, lane 4) were the positive controls for western blot.
study) in Adamts13−/− mice (Figure 3) and up to 42 weeks in wildtype mice (Balb/c) (unpublished data).20 The data demonstrate that a single injection of lentiviral vectors either at E8 or E14, encoding full-length ADAMTS13 and truncated variant into amniotic space or yolk sac vessels, results in long-term expression of ADAMTS13 or variant protease at levels that may be therapeutic.
The expressed full-length ADAMTS13 and truncated variant altered the distribution of vWF multimers To determine whether the expressed ADAMTS13 and MDTCS in mice were able to alter the distribution of circulating vWF 36
multimers in mice, we assessed the plasma multimers by agarose gel electrophoresis followed by western blot. In addition, we also determined the ratio of vWF collagen–binding activity to antigen and multimer analyses. The binding affinity to immobilized human type III collagen is proportional to the size of vWF multimers.21 The high-molecular weight vWF multimers were present in plasma of Adamts13−/− mice or Adamts13−/− mice expressing GFP, but markedly reduced in plasma of mice expressing fulllength ADAMTS13 and MDTCS (both injected at E8 and E14) (Figure 4a). These results were consistent with the significantly reduced mean ratios of plasma vWF collagen–binding activity to antigen by the quantitative measurement in mice expressing full-length ADAMTS13 and MDTCS were detected regardless of the time vectors administered (Figure 4b). These data suggest a substantial alteration of vWF multimer distribution has occurred in the murine plasma containing even low levels of the expressed recombinant ADAMTS13 and MDTCS proteases in circulation.
The expressed human ADAMTS13 and MDTCS offered systemic protection against arterial thrombosis To determine whether the expressed ADAMTS13 and MDTCS proteases were biologically functional in preventing arterial thrombosis, we performed the ferric chloride–induced carotid arterial thrombosis assay. This assay was performed in a blind fashion (the operator was blinded to the genotype of the mice). www.moleculartherapy.org vol. 17 no. 1 jan. 2009
© The American Society of Gene Therapy
a
Potential Gene Therapy for Hereditary TTP
100
E8
MDTCS
−/−
13
-A
3
A1
FL
S
TC
MD
P
GF
13
-A
FL
S
TC
MD
HMW
FL-AD13
80 Protease activity (%)
a
AD13−/−
60 40 20 0 4
8
12
16
20
24
b
FL-AD13
b
AD13−/−
30 20
10
4
6
4
5
8
10
1.4
*P < 0.05 **P < 0.01
The carotid arterial occlusion times in the mice receiving vector at E8 and expressing ADAMTS13 and MDTCS were 9.0 ± 0.6, and 25.2 ± 3.2 minutes, respectively, which were significantly prolonged compared to those of the Adamts13−/− mice expressing eGFP alone (5.6 ± 0.5 minutes) (P < 0.01) (Figure 5a). Similar results were obtained in mice receiving vectors at E14 and expressing ADAMTS13 and MDTCS (Figure 5c). The markedly prolonged carotid arterial occlusion times in mice expressing MDTCS were consistent with higher levels of proteolytic activity (Figure 3). No statistical difference in carotid arterial occlusion times was observed between the mice expressing human ADAMTS13 and wild-type (C57BL/6 strain) mice (8.3 ± 0.4 minutes) (P > 0.05) (Figure 5a,c). In addition, the numbers of occluded arteries at 30 minutes in the mice expressing MDTCS (at E8 and at E14) were ~30%, significantly lower than that in the wild-type mice and the mice expressing ADAMTS13, consistent with the possibility of over correction of Adamts13 deficiency by human MDTCS in these mice or the ectopic expression of the transgene that may be more efficacious for anti-thrombosis. Nevertheless, these results demonstrate that the expressed human full-length ADAMTS13 (although only ~5–7%) and truncated variant (MDTCS) (~10–70%) are biologically functional in vivo. Molecular Therapy vol. 17 no. 1 jan. 2009
1 * 0.8
*
* **
0.6 0.4 0.2
12
Weeks after injection
Figure 3 Proteolytic activity in mouse plasma after in utero injection of lentiviral vectors. Persistent plasma levels of proteolytic activity was detected in mice receiving vector encoding full-length ADAMTS13 (FL-AD13) and variant (MDTCS) at (a) E8 and at (b) E14 by a FRETSvWF73 assay. No detectable proteolytic activity was detected in plasmas of mice receiving a vector encoding eGFP alone (AD13−/−). The normal human plasma (NHP) was used for the reference and arbitrarily defined as 100%.
6
E14 injected
1.2 Ratio of vWF activity to antigen
Protease activity (%)
E14
MDTCS
0
3
E8 injected
50 40
2
1
Weeks after injection
0 GFP
FL-A13
MDTCS
Figure 4 vWF multimer in mice receiving lentiviral vector. (a) vWF multimer distribution of the pooled plasma from Adamts13−/− mice (n = 10, lane 1), Adamts13−/− mice expressing eGFP (n = 10, lane 4), full-length ADAMTS13 (n = 10, lanes 2 and 5) and MDTCS (n = 10, lanes 3 and 6) at E8 and E14, respectively. All plasmas were collected at the age of 6 weeks. The areas of high-molecular weight vWF multimers are indicated. (b) Ratios of vWF collagen–binding activity to antigen in mice receiving vectors at E8 (closed bars) and E14 (open bars) encoding eGFP alone (GFP), full-length ADAMTS13 (FL-AD13) and variant (MDTCS). The ratio in Adamts13−/− mice expressing eGFP alone was defined as 1. The means and standard deviations of the means were shown. One star and double stars indicate statistically significant (P < 0.05) and very significant (P < 0.01) difference, respectively, existing between the experimental group and the control.
DISCUSSION This study reports the successful long-term correction of the prothrombotic phenotypes of Adamts13−/− mice by early gestational gene transfer. Recombinant human ADAMTS13 has been shown to be able to cleave murine vWF in vivo and offers systemic protection against ferric chloride–induced arterial and venous thrombosis after tail vein injection.15 To develop gene therapy–based treatment for hereditary TTP and gain more insight into the structure–function relationship of human ADAMTS13 protease in vivo, human full-length ADAMTS13 and the carboxyl-terminal truncated variant (MDTCS) a well characterized previously,4,5,22 were expressed in Adamts13-null mice by in utero administration of lentiviral vectors. The transgene product (eGFP) after intra-amniotic injection (at E8) of lentiviral vectors was primarily localized to the parenchyma of the kidneys, large vessels of the 37
© The American Society of Gene Therapy
Potential Gene Therapy for Hereditary TTP
c
P < 0.01 30
**
25 20 P < 0.05
15
*
10 5
P < 0.001 30
Occlusion time (min)
0
**
25 20 15
P < 0.05
10
*
5
d
100
80
% Occluded at 30 min
80
60 40 20
− −/
FL
60 40 20
S
13 D
TC D
M
-A FL
13
−/
−
T
S
W
D
TC D
-A FL
M
− −/
13 AD
W
13
0
T
0
-A D 13 M D TC S
T W
100
AD 13
− −/
-A D 13 M D TC S
FL
AD 13
b % Occluded at 30 min
W
T
0
AD
Occlusion time (min)
a
Figure 5 Ferric chloride–induced arterial thrombosis in mice receiving lentiviral vectors. The time to occlude the carotid artery and the number of occluded arteries at 30 minutes in wild-type mice (WT) and Adamts13−/− mice at 24 week after receiving (a, b) intra-amniotic injection at E8 or (c, d) intravascul bp ar injection at E14 of vectors encoding eGFP alone (AD13−/−), full-length ADAMTS13 (FL-A13) and variant (MDTCS), respectively. The Student’s t-test was performed to determine the statistical significance between the control and each experimental group.
heart, and skin (Figure 1a). This is in contrast with the localization of endogenous ADAMTS13 which is detected in hepatic stellate cells,23–25 endothelial cells,26,27 megakaryocytes/platelets.28,29 The predominant expression of ADAMTS13 in renal parenchyma (tubules and glomeruli) by early embryonic injection of lentiviral vector may be important for preventing renal failure in patients with TTP because of thrombosis in kidney microcirculation. The distribution of the transgene expression is largely related to the timing of vector administration and the contact of specific tissues with the amniotic fluid. The amniotic space primarily contacts epithelial tissues at this early gestational time point,17 but is not entirely restricted to epithelium due to exposure of some mesodermal tissue before completion of gastrulation in the primitive streak. There may also be leakage of vector to the extracelomic cavity (such as the expression of transgenes in pancreas) (data not shown), which contacts mesodermal progenitors. When the same vectors were injected via a yolk sac vessel at E14, however, the transgene expression was rather limited to the hepatocytes in the liver (Figure 1b). Thus, the distribution of transgenes observed in this study is consistent with the developmental stage at which the injections were performed and those that we have previously reported.17,18 Despite the cell origin, the synthesized human full-length ADAMTS13 and MDTCS proteins were secreted into the blood stream as demonstrated by the presence of ADAMTS13 (~195K) and MDTCS (~95K) proteins, respectively, in mouse plasma with their expected molecular masses on a SDS-polyacrylamide gel (Figure 2a,b). Although it remains to be determined how such a large protein synthesized from various cells gets into the blood 38
stream, the secreted recombinant full-length ADAMTS13 and MDTCS in murine plasma were proteolytically active toward two highly specific peptide substrates FRETS-vWF73 (Figure 3) and GST-vWF73 (data not shown). This elevated protease activity in murine plasma persisted at least for the length of this study, i.e., 12–24 weeks in the Adamts13−/− mice (Figure 3) and 42 weeks in the wild-type mice observed (our unpublished data).20 Although low levels of plasma protease activity were transduced in mice with human full-length ADAMTS13 (5–7% of normal human plasma), these levels of expression were sufficient to correct the prothrombotic phenotypes. This was indicated by a marked decrease in the ratio of plasma vWF collagen–binding activity to antigen (Figure 4) which is indicative of altered multimer distribution, and a significant prolongation of ferric chloride– induced carotid occlusion times compared to Adamts13 null mice expressing eGFP alone (Figure 5), comparable to the occlusion times of the wild-type mice. These results are also consistent with the clinical data showing that ~5–10% of plasma protease activity may be sufficient to induce clinical remission and maintain patients with TTP symptom-free.16 The lower levels of plasma proteolytic activity in mice expressing ADAMTS13 than in those expressing MDTCS, as determined by cleavage of the peptide substrate may be in part due to lower transduction efficiency (due to longer construct) and perhaps greater instability of ADAMTS13 than MDTCS in murine plasma. The half-life of recombinant human full-length ADAMTS13 and MDTCS in mice was determined to be 15 and 22 minutes (Supplementary Figure S1), respectively, similar to what has been reported for full-length human ADAMTS13 in mice in the literature.15 Consistent with the notion of lower transduction efficiency was that the positive immunofluorescent signals in various tissues of the mice expressing full-length ADAMTS13 were much fewer than those expressing MDTCS as determined by fluorescent microscopy (data not shown). An unanticipated finding was our observation that the truncated MDTCS variant that was previously found to be significantly impaired in proteolytic activity toward plasma-derived vWF under vortex-induced fluid shear stress in vitro was capable of cleaving murine vWF efficiently in vivo (Figure 4) and protecting mice from the ferric chloride–induced arterial thrombosis (Figure 5). The explanation for why the truncated MDTCS remained highly active in processing UL-vWF in vivo is not known. Dong et al. using a parallel flow chamber assay also observed such “hyperactivity” of human MDTCS in proteolytic processing of cell-bound UL-vWF.30 It is possible that the carboxyl terminus of ADAMTS13 that is required for proteolytic cleavage of soluble vWF may not necessarily important for cleavage of newly secreted UL-vWF bound on endothelial cells in vivo. Alternatively, the proteolytic activity of MDTCS may be rescued by the constant arterial shear stress and potential cofactors present in plasma such as factor VIII,31 platelets,32,33 and possibly endothelial cells. Nevertheless, the detection of biological activity of MDTCS effectively reducing vWF multimer sizes in vivo may be important for a rational design of an adeno-associated viral vector for a better safety profile because adeno-associated viral vector has a limited capacity of ~4.3 kb including promoter and other regulatory elements. The MDTCS (~2.4 kb) will fit into the adeno-associated viral vector cassette. www.moleculartherapy.org vol. 17 no. 1 jan. 2009
© The American Society of Gene Therapy
It will be important in the future to demonstrate the efficacy of this approach in mice susceptible to developing TTP-like syndrome. Complete null mutations of Adamts13 in other strains of mice such as those used in this study exhibit prothrombotic phenotypes as evidenced by an accumulation of high-molecular weight vWF multimers in plasma upon stimulation14,10 and an enhanced platelet aggregation and thrombus formation induced by ferric chloride (Figure 5) and by calcium ionophore in vivo.14,15 Therefore, our model should provide reasonable experimental evidence of therapeutic efficacy of the exogenous human ADAMTS13 and MDTCS proteins if the plasma vWF multimer sizes are reduced and the formation of arterial thrombosis are attenuated or prevented. The success in correcting the prothrombotic phenotype in mice by in utero gene transfer of lentiviral vectors encoding either wild-type ADAMTS13 or MDTCS variant supports the feasibility of developing a novel gene therapy–based treatment for hereditary TTP in humans. Although our study is proof-in-principle of this strategy, there are a number of hurdles that would need to be overcome before any clinical application of this approach. First, there are obvious concerns about the fetal loss, the potential for insertional mutagenesis,34 developmental effects, and the potential for germ-line alteration35 that exists for lentiviral vector–based approaches. Although greater tissue specificity and safety can likely be accomplished by the use of tissue specific promoters, or regulated transgene expression, safer gene transfer techniques will need to be developed to alleviate these concerns. The second major impediment is that stage for stage, the timing of our early gestational injections around E8 in the mouse, corresponds to 21–35 days gestation in human development,36 a time in pregnancy that precedes current capabilities for prenatal diagnosis. On the other hand, E14 corresponds to early second trimester that is well within the window for prenatal diagnosis and intervention. In the foreseeable future, prenatal diagnosis may allow diagnosis of genetic disorders within an earlier timeframe. Finally, this model system may be useful for experimental screening of various constructs for efficacy in treatment of TTP associated with ADAMTS13 deficiency, for instance, if postnatal epidermal based secretory strategies are contemplated.37
MATERIALS AND METHODS Construction and preparation of recombinant lentiviral vectors. The ZHK construct is a self-inactivating, replication-incompetent HIV-1based lentiviral vector that has previously been described.17 The transgene cassette was composed of the human cytomegalovirus (CMV) immediate early promoter driving eGFP expression and human ADAMTS13 or the variant truncated after the spacer domain (MDTCS) (Figure 6). Bicistronic expression was accomplished by inserting the therapeutic gene downstream and in frame with the eGFP cDNA and TaV sequence, a cis-acting hydrolase element derived from the Thosea asigna virus.38 Vesicular stomatitis virus–G pseudotyped vectors were produced by calcium phosphate–mediated transient transfection of HEK293T cells.39 Briefly, cells cultured in DMEM (GIBCO-Invitrogen, Carlsbad, CA) containing 10% fetal bovine serum (HyClone, Logan, UT), 100 U/ml of penicillin, and 100 mg/ml streptomycin (GIBCO-Invitrogen) were co-transfected with appropriate amounts of the lentiviral vector plasmid, the packaging plasmid, pCMVΔ8.91, and the vesicular stomatitis virus–G expression plasmid, pHCMVG.40 Approximately 16 hours before virus isolation, the media were replaced with similar media minus phenol red. High-titer Molecular Therapy vol. 17 no. 1 jan. 2009
Potential Gene Therapy for Hereditary TTP
Packaging signal R
U5
WPRE-B11
U3 R
FLAP
GFP
CMV
eGFP
FL-AD13
CMV
eGFP
U5
Tev ADAMTS13 (aa 1-1427) Tev MDTCS
CMV
eGFP
MDTCS (aa 1-685)
Figure 6 Schematic representation of the lentiviral vector and the encoded human full-length ADAMTS13 and C-terminal truncated variant. Human ADAMTS13 (amino acid residues 1–1427, ADAMTS13) and a variant truncated after the spacer domain (amino acid residues 1–685, MDTCS) were cloned into a self-inactivating HIV-1-based vector to form ZHK-CMV-eGFP-Tav3-ADAMTS13 (7,953 bp) and ZHK-CMV-eGFPTav3-MDTCS (5,726 bp). These vectors contain a cytomegalovirus (CMV) promoter, an enhanced green fluorescent protein (eGFP) reporter gene and Tav sequence. The rev response element (RRE) and the Woodchuck hepatitis virus posttranscriptional regulatory element (WPRE) enhance transcription and expression of the transgenes as indicated above the schematic vector structures.
vectors were prepared by concentrating viral supernatants with ultracentrifugation, then aliquoted and stored at −80 °C. The virus stocks were titered on HEK293T cells by quantifying eGFP positive cells, ranging from 108 to 1010 transducing units (TU)/ml. In utero administration of lentiviral vectors. Adamts13−/− mice
(C57BL/6J/129Sv) were generated by crossing Adamts13B/129+/− mice13 with C57BL/6 mice (Taconic, Rensselaer, NY) to generate Adamts13+/− and Adamts13−/− mice. The mice were crossed for at least four generations before experimentation. Time-dated pregnant mice at postcoital days 8 (E8) and 14 (E14) were anesthetized by isoflurane inhalation (3.5% (vol/ vol) for induction and 2% (vol/vol) for maintenance). The intra-amniotic17,18 and intravascular injections41,19 were performed at E8 and E14, respectively, as previously described. Stereoscopic fluorescent microscopy. Newborn mice were killed and vari-
ous organs were directly placed under a stereoscopic fluorescent microscope (MZ16FA; Leica, Heerburg, Switzerland) to visualize the reporter eGFP expression.
Immunofluorescent microscopy. Various mouse tissues were sectioned
with a cryostat after being frozen in TissueTek OCT-embedding medium (Miles, Elkhart, Indiana). The cryosections were fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS). After being permeabilized with 1% Triton X-100, the sections were blocked with 2% fetal bovine serum (Invitrogen, Carlsbad, CA) in PBS, the sections were incubated anti-GFP IgG (1:100) in 2% fetal bovine serum in PBS at 4 °C overnight. The bound antibody was detected by incubation with Cy2conjugated anti-rabbit (1:200) (Jackson ImmunoResearch, West Grove, PA). The nuclei were stained with 4a,6-diamidin-2-phenylindole in the mounting medium (Vector Laboratories, Burlingame, CA). The images were taken under a Leica Inverted fluorescent microscope (Leica, Wetzlar, Germany).
Immunoperoxidase staining. Tissue specimens were fixed in 10% buffered formalin solution and embedded in paraffin. Sections were prepared using a paraffin microtome and deparaffinized by a standard protocol. After being blocked with goat serum (1:10 dilution), the slides were incubated with rabbit anti-GFP IgG (1:200) (Molecular Probes, Eugene, OR), followed by an incubation with biotinylated goat anti-rabbit IgG, avidin– peroxidase and peroxidase substrate (Vector Laboratories) as described previously.18 The cell nuclei were lightly stained with Harris hematoxylin.
39
© The American Society of Gene Therapy
Potential Gene Therapy for Hereditary TTP
Blood collection. Blood was collected from retinal sinus via capillary tube using 3.2% sodium citrate as anticoagulant at 4, 8, 12, 16, 20, and 24 weeks after the injection of lentiviral vectors. After centrifugation at 1,000g for 5 minutes in a microcentrifuge, plasma was aspirated, aliquoted, and frozen at −80 °C until assay. ADAMTS13 activity. The proteolytic activity in mouse plasma was determined by FRETS-vWF73 assay42,23 as described previously. To estimate the concentration of recombinant ADAMTS13 and variant in murine plasma, pooled normal human plasma was used as a reference and defined as 100% of activity. Immunoprecipitation and western blot. To detect full-length ADAMTS13
in mouse plasma, 100 μl of mouse plasma pooled from five mice expressing eGFP (controls) and ADAMTS13 at 5–6 weeks of age or normal human plasma was diluted 1:2 with 100 μl of PBS and incubated with 20 μl each of protein G/A Sepharose 4B (BD Biosciences, San Jose, CA) at room temperature for 1 hour. After mouse IgG was removed, 100 μg of purified human IgG autoantibody against ADAMTS13 isolated from a patient with acquired TTP that recognizes both N- and C-termini of human ADAMTS13 (unpublished data) was added and incubated at 25 °C for 30 minutes. The antibody–antigen complexes were pulled down by incubation with 15 μl of protein G/A Sepharose 4B for 60 minutes. After washing the Sepharose 4B beads once with PBS, the bound immune complexes were eluted from the beads by boiling them at 100 °C for 5 minutes and detected by western blot with rabbit anti-ADAMTS13 IgG (kindly provided by Dr Fritz Scheiflinger, Baxter Biosciences, Orth, Austria) as described previously.26,27 Alternatively, to detect MDTCS in mouse plasma, the immunoprecipitation was modified in such to avoid nonspecific protein bands obscuring visualization of the protein of interest (~97 kDa). Pooled plasma (100 μl) from five mice expressing eGFP and MDTCS was incubated with 50 μl of the same patient IgG autoantibody against ADAMTS13 (~100 µg IgG) that was covalently immobilized onto Affigel10 (Bio-Rad, Hercules, CA) (2 mg/ml beads) at 4 °C overnight. The bound recombinant human MDTCS was eluted from the beads by adding 40 μl of 0.1 mol/l glycine-HCl, pH 2.5. The MDTCS protein was then detected by western blot with mouse anti-Dis IgG (1:1,250), followed by incubation with IRDye CW800-labeled anti-mouse IgG (1:12,500) and imaging on an Odyssey Imaging System (LI-COR, Lincoln, Nebraska).
vWF antigen assay. Plasma vWF antigen was determined by an ELISA assay using two different rabbit anti-vWF antibodies.43 Briefly, plasma vWF was captured by rabbit anti-vWF IgG (DAKO, Carpinteria, CA) (10 µg/ml) immobilized on microtiter plate and detected by peroxidase-conjugated rabbit anti-vWF IgG (1:5,000) (DAKO), followed by incubation with ophenylenediamine, (OPD, Sigma, St Louis, MO)-H2O2. To stop the reaction, 100 µl of 1.5 N H2SO4 was added. The plate was read at 492/620 nm on SpectraMax 190 ELISA reader (Molecular Devices, Union City, CA). vWF collagen–binding activity. Citrated plasma (10 µl) was diluted 1:20
(vol/vol) with PBS and 100 µl was added to the wells coated with human collagen type III (10 µg/ml) in duplicate. After incubation at 25 °C for 1 hour, the bound vWF was detected by peroxidase-conjugated rabbit antivWF IgG (1:3,000) and substrate OPD-H2O2 as described previously.44,45 Pooled normal murine plasma was used in as a reference. Plasma vWF multimers. The vWF multimers were performed according to
a method described by Banno et al.14 with modifications. Ten microliters of citrated mouse plasma was denatured by heating at 60 °C for 20 minutes in 190 μl of 50 mmol/l Tris-HCl, pH 6.5, containing 10% (wt/vol) SDS, 42% (wt/vol) glycerol, 2 mmol/l EDTA, and 0.02% bromphenal blue. The denatured sample (30 μl) was fractionated in 1% (wt/vol) SeaKem HGT agarose (Cambrex, East Rutherford, NJ) gel by electrophoresis at 15 mA for 2.5 hours on ice. After being transferred onto nitrocellulose membrane (BioRad) at 100 mA for 60 minutes on a Hoefer TE77 apparatus, the membrane
40
was blocked by 1% (wt/vol) casein in TBST for 30 minutes and incubated with rabbit anti-vWF IgG (A0082 DAKO) (1:1,500) in 1% (wt/vol) casein TBST overnight, followed by IRDye800 fluorescent-conjugated goat antirabbit IgG (1:12,500), 25 °C for 1 hour. The membrane was washed three times with TBST, once with PBS and scanned in an Odyssey Imaging System at intensity of 5.0. Carotid arterial thrombosis. The mice at the age of 24 weeks (at the end of observation for long-term expression) were anesthetized with intraperitoneal injection of 10 mg/ml Nembutal (0.1 ml/10 g) and the right carotid artery was exposed by blunt dissection. A Doppler flow probe (Model 0.5VB, Transonic Systems, Ithaca, NY) was placed around the artery.46,47 Thrombosis was induced in the exposed carotid artery by applying a piece of filter paper (1 × 2 mm) saturated with 10% (vol/vol) ferric chloride to the adventitia for 1 minute. The field was flushed with PBS, and the blood flow was monitored for 30 minutes. The time until initial complete occlusion and the presence or absence of arterial occlusion at 30 minutes was recorded in the experimental and control mice. Statistical analysis. Student’s t-test was used for continuous variables and analysis of variance and χ2-tests were used to determine the differences among various groups. P values
