Fetal gene therapy of α-thalassemia in a mouse model

Abstract

Fetuses with homozygous α-thalassemia usually die at the third trimester of pregnancy or soon after birth. Hence, the disease could potentially be a target for fetal gene therapy. We have previously established a mouse model of α-thalassemia. These mice mimic the human α-thalassemic conditions and can be used as preclinical models for fetal gene therapy. We tested a lentiviral vector containing the HS 2, 3, and 4 of the β-LCR, a central polypurine tract element, and the β-globin gene promoter directing either the EGFP or the human α-globin gene. We showed that the GFP expression was erythroid-specific and detected in BFU-E colonies and the erythroid progenies of CFU-GEMM. For in utero gene delivery, we did yolk sac vessel injection at midgestation of mouse embryos. The recipient mice were analyzed after birth for human α-globin gene expression. In the newborn, human α-globin gene expression was detected in the liver, spleen, and peripheral blood. The human α-globin gene expression was at the peak at 3–4 months, when it reached 20% in some recipients. However, the expression declined at 7 months. Colony-forming assays in these mice showed low abundance of the transduced human α-globin gene in their BFU-E and CFU-GEMM and the lack of its transcript. Thus, lentiviral vectors can be an effective vehicle for delivering the human α-globin gene into erythroid cells in utero, but, in the mouse model, delivery at late midgestation could not transduce hematopoietic stem cells adequately to sustain gene expression.

The human α-globin genes are duplicated, and four copies of α-globin genes are present in the diploid genome. α-Thalassemia is a hereditary disorder caused by deficient or absent production of α-globin. α-Globin gene mutation frequency is high among many populations, and the severe form has the highest prevalence in Southeast Asia. Hydrops fetalis associated with hemoglobin Bart syndrome is caused by complete absence of α-globin and is usually not compatible with postnatal life. Hemoglobin H disease, caused by a deletion and/or mutation affecting three of the α-globin genes, results in hemolytic anemia of variable severity. The prenatal genetic diagnosis for α-thalassemia has been clinically available for many years (1). A few patients with homozygous α-thalassemia have survived by early and regular blood transfusions (2–6), and hemoglobin H disease is usually treated symptomatically.

Recombinant lentiviral vectors have been shown to be effective in transducing nondividing hematopoietic stem cells (7–10). By using lentiviral vectors carrying the β-globin transcription units and ex vivo transduction, therapeutic β-hemoglobin synthesis has been demonstrated in β-thalassemic mice (11)) as well as the antisickling capability of the β-globin variant in a transgenic mouse model of sickle cell disease (12).

Direct in vivo delivery of the therapeutic viral vector has been shown in animal models to be an effective alternative to the ex vivo approach (13, 14). Particularly, direct in utero viral vector transfer has been demonstrated to result in widespread transduction and long-term correction of transduced genes in animal models of human genetic diseases such as lysosomal storage disease, Crigler–Najjar disease, and Duchenne muscular dystrophy (15–17).

In this study, we investigated the efficacy of direct in utero delivery of a lentiviral-based human a-globin gene to a mouse model of α-thalassemia. We demonstrated erythroid specific expression of the transduced human α-globin gene and relatively high levels of expression of the human α-globin gene in mice receiving the lentiviral vector by yolk sac vessel injection at midgestation. However, the expression decreased to low levels on long-term follow up.

Results

Lentiviral Vector Construction.

The lentiviral vector used in this study was derived from a TNS9 vector (11), which contains an extended β-promoter, β-proximal enhancers, and genomic fragments of the human β-globin control region, HS2, HS3, and HS4. This vector has been shown in ex vivo transduction experiments to be effective in transferring the therapeutic human β-globin gene into murine hematopoietic stem cells (11, 18–20). All elements in the original vector remained except for the replacement of the β-globin gene with either the human α-globin gene or the cDNA-encoding GFP (Fig. 1A). Southern blot analysis of genomic DNA from infected mouse erythroleukemia (MEL) cells showed a single band corresponding to the intact proviral vector (Fig. 1B).

Fig. 1.

Structures and integration of the lentiviral vectors. (A) Exons and introns of the human α-globin gene are represented by filled and open boxes respectively and GFP coding region by the hatched box. The splice donor (SD) and acceptor (SA), the packaging sequence (ψ), and rev-response element (RRE), DNase hypersensitive sites (HS), human β-globin promoter (P), and 3′ β-globin enhancer (E) are derived from the lentiviral vector, TNS9 [11]. The inverted triangle indicates the 3′ LTR deletion and cppt central polypurine tract. (B) Southern blot analysis of transduced MEL cells. Genomic DNA of MEL cells infected with dANS9-cppt-hα vector was digested with restriction enzyme SacI and hybridized with the human α-globin probe. The upper band represents the expected size for the proviral vector, and the lower band was due to cross hybridization with the endogenous mouse α-globin gene. The λ/HindIII DNA molecular markers are indicated in kilobases.

Erythroid-Specific Expression of GFP.

To evaluate whether gene expression from these lentiviral vectors are erythroid-specific, K562 and MEL cells were infected with the viral supernatant of dANS9-cppt-egfp. GFP expression was detected when these cells were induced toward erythroid differentiation (Fig. 2A), whereas no expression was detected in infected 293 cells (data not shown). When this vector was used to infect primary murine bone marrow cells, GFP expression was detected in BFU-E colonies or erythroid lineage of CFU-GEMM colonies, but not in cells of other lineages derived from the same progenitors (Fig. 2B).

Fig. 2.

Expression of GFP in vitro transduction studies. (A) Erythroid cell lines K562 and MEL were infected with the lentiviral vector dANS9-cppt-egfp followed by induction for 5 days with HMBA for MEL cells and Hemin for K562 cells. (B) Mouse bone marrow cells were infected with the lentiviral vector dANS9-cppt-egfp and cultured in complete methylcellulose medium supplemented with IL-3, IL-6, SCF, and erythropoietin for 10 days. Green fluorescence was detected under fluorescent microscope. GFP-positive cells were observed in CFU-Emix (Left) and BFU-E (Right). Images with UV light (Upper) and visible and UV light (Lower) are shown. The arrows indicate the nonerythroid colony that was negative for GFP.

Direct Delivery of the Lentiviral Vectors into Fetuses by Yolk Sac Vessel Injection.

Direct administration of the therapeutic viral vectors into the fetus can eliminate multiple manipulating steps associated with ex vivo gene transfer. Therefore, a modified yolk sac vessel injection protocol was used in this study to deliver the viral vector systematically into the fetuses (21, 22). By injecting Trypan blue dye as tracer, we could see blue color appearing within seconds in the fetal circulation. Fig. 3 shows the site of injection (Left), followed immediately by the appearance of the dye in the yolk sac vessels (Center). They can be distinguished from the uterine wall vessels, which were not stained blue. In addition, the dye was concentrated in the liver as shown by the high dye intensity in its location (Right). This is important for targeting hematopoietic progenitors because the fetal liver is a major site for hematopoietic stem cell homing and development. In the mouse, fetal liver erythropoiesis starts from day 11, progressing to around day 14, when the erythroblastic islands are established (23). Although it would be preferable to target the fetal liver during the early part of this period when it is relatively richer in hematopoietic stem cells (24), we injected on day 14.5 because of technical difficulties and the high mortality associated with earlier injections.

Fig. 3.

In utero delivery of lentiviral vector by yolk vessel injection. (Left) Needle insertion site. (Center) Dye in yolk sac vessel after injection. ysv, yolk sac vessel; uwv, uterine wall vessel. (Right) Concentration of the dye in fetal liver after injection.

Efficacy of Gene Transfer.

To investigate the effectiveness of gene transfer of the dANS9-cppt-egfp and dANS9-cppt-ha vectors via yolk sac vessel injection, we analyzed the transgene expression in newborn recipient mice. GFP expression was detected in spleen for dANS9-cppt-egfp (Fig. 4A) and human a-globin specific transcripts in liver for dANS9-cppt-hα (Fig. 4B).

Fig. 4.

Gene expression in the newborn mice after in utero lentiviral injection. (A) GFP expression in frozen section of the spleen after injection of dANS9-cppt-egfp. (B) Human α-globin RNA detection in the liver. Digestion of the 350-bp human α-globin RT-PCR product with Pml 1 yields 180- and 170-bp bands and with HindIII 230/120-bp bands. The left side shows injected newborn mouse liver, and the right side shows human fetal liver control. U, undigested; H, HindIII; P, Pml I.

Long-Term Expression of Erythroid-Specific Lentiviral Vectors.

Primer extension analysis was performed to measure the transgene expression relative to the endogenous mouse α-globin gene expression in peripheral blood at different time points after birth (Fig. 5A). Expression of the transduced human α-globin gene was detectable at ≈70 days and reached peak levels at 130 days after birth, indicating the expansion of the targeted erythroid progenitors. The human α-globin transcript was expressed at a level equal to 20% of the mouse α-globin transcript in one mouse. However, in all of the mice, expression declined to the low level of 5% or less in recipient's peripheral blood 7 months after birth (Fig. 5B).

Fig. 5.

Primer extension analysis of human α-globin gene expression in three recipient mice. Total RNA was prepared from peripheral blood at 70 days (lane a), 90 days (lane b), 130 days (lane c), and 210 days (lane d) after birth. c1 and c2 are positive (human fetal liver) and negative (uninjected mouse) controls respectively. (A) Autoradiography. (B) Human α globin mRNA relative to total human and mouse α-globin mRNA determined by PhosphorImager.

Possible Reason for the Decline in Transgene Expression.

We harvested bone marrows from two of the mice that had shown a decline in α-globin gene expression and cultured BFU-E and CFU-Emix from them. Fifty-five individual colonies were analyzed for the human α-globin DNA by PCR and for the presence of human and mouse α-globin mRNA by RT-PCR. Twenty-four of these colonies were also cultured in the presence of 5-azacytidine (Table 1). Human α-globin DNA was found in only 6 of these 55 colonies, and α-globin transcripts were found in none, with or without 5-azacytidine. In contrast, mouse β-globin mRNA was detected in 49 of these colonies, showing that the RNA prepared from them were of good quality. Hence, the decline in gene expression could be due to the loss of cells that contained the human α-globin transgene and to the silencing or level too low to be detected in the few that retained the transgene.

Table 1.

Number of CFUs assayed for human α-globin transgene and mRNA and mouse mRNA

5-aza	n	h DNA	h RNA	m RNA
+	24	2	0	22
−	31	4	0	27
Total	55	6	0	49

Discussion

In this study, we investigated the approach of intrauterine therapy for α-thalassemia in an α-globin gene knockout mouse model that we have constructed (25). Homozygous α-thalassemia appears to be a good candidate for intrauterine gene therapy because prenatal diagnosis can be made by DNA analysis early in pregnancy, and the homozygously affected fetuses often survive up to the third trimester of pregnancy or to birth. Although several newborns with this disease survived when they received transfusion prenatally or immediately after birth, the anemia is invariably severe, and they invariable require regular transfusion beginning from birth. Thus, the disease is more severe than β-thalassemia, where the clinical manifestation is much more variable, with some patients having the less severe intermediate form. Even those who require transfusion may not need it until a few months or a few years of age. Successful intrauterine gene therapy in α-thalassemia would therefore obviate subsequent transfusion.

Using a dye as a marker, we showed that injection into the yolk sac vein resulted in concentration of the dye in the liver. Van der Wegen et al. (26) reported the successful treatment of UDP-glucuronosyltransferase deficiency in a rat model of Crigler–Najjar disease by using a liver-specific lentiviral vector. Targeting the liver is a distinct advantage for treating globin disorders because the fetal liver is a hematopoietic organ. However, some vectors will also be expected to reach the systemic circulation. Hence, the use of erythroid-specific promoters and enhancers is desirable. Lentiviral vectors were chosen to deliver the GFP and the α-globin genes because they could transduce hematopoietic cells efficiently and were successful in treating mouse models of β-thalassemia and sickle cell anemia by ex vivo transduction of hematopoietic cells. Indeed, we also found that our vector specifically expressed GFP in erythroid cell lines and in erythroid colonies cultured from mouse bone marrow cells.

The α-globin gene we used was controlled by the β-LCR because such a construct has been shown to express the α-globin efficiently in transgenic mice (27). Control by the β-globin elements would also be expected to maintain gene expression into postnatal life. Intrauterine injection resulted in expression of the α-globin gene at birth and reached a peak of up to 20% in some mice at ≈3–4 months. However, the level of expression declined to <5% at approximately the seventh month. The decline could be due to several factors. Very few erythroid colonies cultured from the bone marrow of the mice after their α-globin expression has declined contained the α-globin transgene. This suggests that the decline may be due to the loss of the transgene as a result of the inadequate transduction of hematopoietic stem cells, and most of the transduced cells were later progenitors. Because of technical difficulty and the high mortality rate, delivery of the lentiviral vectors close to day 14.5 of gestation and not to day 11 may account for the transduction of the later progenitor cells. Unlike effective fetal treatment of metabolic diseases, where the liver cell could be targeted during most of the duration of pregnancy, treatment of hematopoietic disorders may have a much narrower window for effective hematopoietic stem cell targeting. Our inability to detect human α-globin mRNA in the few human α-globin gene-containing colonies may be due to the low level of the transcripts or to gene silencing that cannot be reactivated by demethylating agents.

In summary, we demonstrated the erythroid-specific expression of the human α-globin gene after the lentiviral-based vector delivery at midgestation by yolk sac vessel injection. However, the expression declined at 7 months. The decline may be due to the inability to transduce early hematopoietic stem cells by delivery at 14.5 days of gestation or to gene silencing. Larger-animal models in which the fetal circulation can be accessed earlier may well be used to test the efficacy of in vivo lentiviral transduction of hematopoietic stem cells.

Materials and Methods

Construction of the Lentiviral Vectors.

The human α-globin gene fragment spanning NcoI and PstI sites was used to replace the β-globin gene in the lentiviral vector TNS9 (11). The central polypurine tract (cppt) fragment was amplified from the pol region of the packaging vector pCMV-Δ8.9 (14) and inserted into ClaI and HapI sites downstream of the 3′ β-enhancer. The primers used for the cppt fragment amplification were as follow: cppt-5-HpaI, TCGCGTTAACTTTTAAAAGAAAAGGGGGG and cppt-3-ClaI, AAGCTTATCGATAAAATTTTGAATTTTTGTAATTTG. The orientation of cppt was checked by sequencing. Subsequently, the human α-globin gene in the resulting vectors, dANS9-cppt-hα, was replaced with the EGFP gene fragment from pIRES-EGFP (Clontech, Mountain View, CA) to make the control vector dANS9-cppt-egfp.

Production and Purification of Lentiviral Vector.

The vector, dANS9-cppt-hα or dANS9-cppt-egfp, was cotransfected with pCMV ΔR8.9 (14) and pMD.G (28) into human embryonic kidney cells 293FT as described (8). Briefly, 6 × 10⁸ of 293FT cells in 750 ml of DMEM containing 10% FBS (D10F) were plated into a polylysine-precoated cell factory unit (Nalge Nunc International, Rochester, NY) and transfected the next day with 580 μg of dANS9-cppt-hα or dANS9-cppt-egfp, 430 μg of pCMVΔR8.9, and 145 μg of pMD.G in a 5% CO₂ incubator at 37°C for 24 h. After the transfection medium was removed, the cells were washed once with DMEM, and the transfected cells were incubated for another 24 h in fresh D10F supplemented with 10 mM sodium butyrate (Sigma, St. Louis, MO) and 20 mM Hepes. The cells were washed twice with DMEM and incubated with 450 ml of DMEM containing 5% FBS and 20 mM Hepes for virus production. The viral supernatants were collected on three consecutive days, spun at 3,500 rpm [RC-3B; Sorvall (Newtown, CT) at 4°C for 10 min and filtered through a 0.45-μm low-protein-binding filter (Millipore, Bedford, MA)]. The viral supernatant was concentrated by two rounds of ultracentrifugation at 25,000 rpm, 15°C for 100 min in a SW32Ti rotor in Optima L-90K Ultracentrifuge (Beckman, Fullerton, CA). The final viral pellet was resuspended in 1 ml of saline containing 4 μg/ml Polybrene and stored at −80°C until use.

Virus titer was determined by a p24 antigen ELISA, following the manufacturer's instructions (ZeptoMetrix, Buffalo, NY). The number of erythroid-specific infectious viral particles was determined by infecting MEL cells with dANS9-cppt-egfp lentiviral vector, followed by induction for 4–5 days with 5 mM N,N′-hexamethylene bisacetamid (HMBA; Sigma).

Cell Lines and CFU Assay.

MEL and K562 cells were carried in RPMI medium 1640 containing 10% of FCS. For infection, cells were incubated with the viral solution containing 10 μg/ml of Polybrene for 4–6 h, followed by induction for 5 days with HMBA for MEL cells and hemin for K562 cells, respectively. Bone marrow cells were prepared from the α-globin gene knockout mice and infected overnight with the viral solution in the presence of 10 μg/ml Polybrene and plated into complete methylcellulose medium containing IL-3, IL-6, SCF, and erythropoietin (M3434; StemCell Technologies, Vancouver, BC, Canada). The colonies were scored after 10 days of incubation.

CFUs were similarly prepared from the bone marrow of two of the injected mice that had a decline of α-globin expression after 7 months. Half of the cultures was incubated in the presence of 5 μM 5-azacytidine (Sigma). After 12 days, cells from an individual BFU-E or CFU-Emix colony were transferred to a 2-ml tube containing ULTRASPEC RNA solution, and total RNA and DNA were isolated according to the manufacturer's instruction and DNA and RNA determined as described.

Southern Blot Analysis.

Genomic DNA was isolated from infected MEL cells by proteinase K digestion and sodium chloride precipitation. Ten micrograms of DNA were digested with ScaI, electrophoresed on a 1% agarose gel, transferred to a Hybond-n + membrane (Amersham Biosciences, Piscataway, NJ), and hybridized to a digoxigenin-labeled human α-globin probe at 65°C in 0.5M sodium phosphate buffer and 6% SDS. After washing at 65°C, the hybridized products were detected by chemiluminescence by using an antidigoxigenin alkaline phosphatase conjugate and the CSPD substrate (Roche, Indianapolis, IN).

Animal Model and Yolk Sac Vessel Injection.

The α-globin knockout mice were produced as described (25). The heterozygous (−α/αα) females were bred with the homozygous (−α/−α) males. Pregnant mice at day 14.5 of gestation were anesthetized by i.p. injection of 0.1 ml of 2.4% tribromoethanol (Sigma–Aldrich) per 10 g of body weight. A midline laparotomy (1–1.5 cm) was performed to expose horns of the gravid uteri. The yolk sac vessels of individual embryos were visualized under a dissecting microscope. The injection was done by inserting a glass needle (70–80 μm) attaching to a Hamilton microliter syringe. Injection volume was controlled through a PB600–1 repeating dispenser (Hamilton, Reno, NV). Ten to 15 μl of lentiviral vector equivalent to 5 × 10⁵ to 1 × 10⁶ infectious viral particles was injected into each embryo. After injection, the uteri were returned to the abdominal cavity and the abdomen was closed with 4–0 silk suture. The mice were allowed to recover in a warm cage. All animal experiments were carried out according to the institutional guidelines for animal use.

RT-PCR Analysis.

Peripheral blood samples or tissues were lysed in ULTRASPEC RNA solution (Bioteck Laboratories, Houston, TX), and total RNA was isolated, following the manufacturer's instructions. For RT-PCR, cDNA was synthesized from 1 μg of total RNA in 10 μl of RT buffer containing 5 mM dNTP, 10 μM oligo d(T), 10 units of RNase inhibitor, and 2 units of M-MLV reverse transcriptase at 37°C for 50 min, followed by 70°C for 15 min. PCR was carried out by using the following primers: Hα660+: 5′-TAAGGTCGGCGCGCACGCTGGC and Hα1266-: AAGCCAGGAACTTGTCCAGG. The reactions were first denatured at 94°C for 5 min, followed by 40 cycles of 94°C for 30 seconds, 62°C for 30 seconds, and 72°C for 30 seconds. The final extension was carried out at 72°C for 5 min. The PCR products were digested with restriction enzyme Pml I or HindIII and analyzed by electrophoresis on 2% agarose gel.

Primer Extension Analysis.

A primer extension assay was done by using the AMV Reverse Transcription Primer Extension System (Promega, Madison, WI) with γ-[³²P]ATP-labeled primers as the following: mβ-98 5′-AGCAGCCTTCTCAGCATCAG, resulting in a 98-nt band for mouse β-globin; mα-53 5′-TGATGTCTGTTTCTGGGGTTGTG; hα-82 5′-CGTTGGTCTTGTCGGCAGGAAAC. The labeled primers were annealed to 1 μg of total RNA, and the reaction was carried our according the manufacturer's instruction. The intensity of radioactive bands was determined with PhosphorImager analysis.

Abbreviations

cppt

central polypurine tract

MEL

mouse erythroleukemia.