Successful Therapeutic Effect in a Mouse Model of Erythropoietic Protoporphyria by Partial Genetic Correction and ﬂuorescence-Based Selection of Hematopoietic Cells

Gene Therapy (2001) 8, 618–626  2001 Nature Publishing Group All rights reserved 0969-7128/01 $15.00 www.nature.com/gt RESEARCH ARTICLE Successful therapeutic effect in a mouse model of erythropoietic protoporphyria by partial genetic correction and ﬂuorescence-based selection of hematopoietic cells

A Fontanellas1,2, M Mendez1, F Mazurier1, M Cario-Andre´1, S Navarro2, C Ged1, L Taine1,3, FGe´ronimi1, E Richard1, F Moreau-Gaudry1, R Enriquez de Salamanca2 and H de Verneuil1 1Laboratoire de Pathologie Molećulaire et The´rapie Geńique, Universite´ Victor Segalen Bordeaux 2, France; 2Centro de Investigacioń, Hospital 12 de Octubre, Madrid, Spain; and 3Laboratoire de Geńe´tique, CHU Pellegrin, Bordeaux, France

Erythropoietic protoporphyria is characterized clinically by transplantation, the number of fluorescent erythrocytes skin photosensitivity and biochemically by a ferrochelatase decreased from 61% (EPP mice) to 19% for EPP mice deficiency resulting in an excessive accumulation of photore- engrafted with low fluorescent selected BM cells. Absence active protoporphyrin in erythrocytes, plasma and other of skin photosensitivity was observed in mice with less than organs. The availability of the Fechm1Pas/Fechm1Pas murine 20% of fluorescent RBC. A partial phenotypic correction was model allowed us to test a gene therapy protocol to correct found for animals with 20 to 40% of fluorescent RBC. In con- the porphyric phenotype. Gene therapy was performed by ex clusion, a partial correction of bone marrow cells is sufficient vivo transfer of human ferrochelatase cDNA with a retroviral to reverse the porphyric phenotype and restore normal hem- vector to deficient hematopoietic cells, followed by re-injec- atopoiesis. This selection system represents a rapid and tion of the transduced cells with or without selection in the efficient procedure and an excellent alternative to the use of porphyric mouse. Genetically corrected cells were separated potentially harmful gene markers in retroviral vectors. Gene by FACS from deficient ones by the absence of fluorescence Therapy (2001) 8, 618–626. when illuminated under ultraviolet light. Five months after

Keywords: animal model; porphyria; cutaneous photosensitivity; genetic disease; bone marrow; gene transfer; retrovirus

Introduction ure, which necessitates liver transplantation.1,2,4–7 The therapy of EPP is often limited to supportive care and is Erythropoietic protoporphyria (EPP) is a hereditary dis- only partially successful, especially in the severe forms. order caused by a decrease in ferrochelatase activity This monogenic disorder starts early in infancy and rep- (E.C.4.99.1.1). The enzyme is associated with the inner resents a good candidate for gene therapy in severe forms mitochondrial membrane and catalyzes the insertion of 1 since the genetic defect is well characterized at the mol- iron into protoporphyrin to form heme. This defect ecular level. Generally, EPP is an autosomal dominant results in a high accumulation of protoporphyrin in disease, but in some cases it is transmitted in a recessive erythrocytes, plasma, feces and other tissues such as the fashion.8–10 Following the cloning of the human ferrochel- liver and the skin.1,2 Clinically, the disease is charac- atase cDNA, several mutations have been identified terized by cutaneous photosensitivity due to the photore- showing a high molecular heterogeneity.8–13 active properties of protoporphyrin. Light between 360 and 450 nm initiates a chain of reactions in which proto- The pathophysiology of the human disease is still unclear, particularly with respect to the development of porphyrin acts as a photochemical radical. This photosen- 14 sitizing property induces an acute inflamation of the skin, liver failure. Acute hepatic failure is due to the accumu- characterized by burning pain, oedema and often pur- lation of protoporphyrin in the liver. Because the proto- pura.3 Occasionally, a mild microcytic, hypochromic ane- porphyrin mostly originates from the erythroid cells, mia may be observed. Signs of hepatic damage have also bone marrow (BM) transplantation can be a convenient been described, and can vary in severity from elevated therapy for severe cases of EPP. An EPP patient who serum transaminase activity to life-threatening liver fail- underwent allogeneic BM transplantation for leukemia had complete hematologic remission.15 A murine model of EPP, the Fechm1Pas/Fechm1Pas mouse16,17 was used for cell18,19 and gene therapy.18 The replacement of the major Correspondence: Fontanellas, Centro de Ingestigacioń, Hospital part of the deficient BM cells by normal ones was able Universitario 12 de Octubre, Avenida de Cordoba, Km 5.400, 28041, to reduce protoporphyrin accumulation in erythrocytes. Madrid, Spain Received 27 October 2000; accepted 15 January 2001 Also, gene therapy to the bone marrow in the same Absence of porphyria phenotype following partial gene correction A Fontanellas et al 619 model can reverse the protoporphyrin accumulation as well as the photosensitization associated with the porphyria.18 Interestingly, BM transplantation performed in very young animals (3–4 weeks old) can prevent hepatobiliary complications as well as hepatocyte alterations and partially revert protoporphyrin accumulation in the liver.19 However, allogeneic BM transplantation is limited by the need for an HLA-matched donor. In the absence of a suitable donor, autografting of genetically modified hematological cells seems to be an appealing alternative. An important step in view of future gene therapy in humans is developing efficient selection procedures to increase the frequency of genetically corrected cells before auto- Figure 1 (a) Identification of LFSN proviruses by Southern blot analysis logous transplantation. One approach relies on the in FDCP1 transduced and selected cells. Genomic DNA was prepared expression of a cotransfected marker gene. In the recent from normal, untransduced and transduced G418 selected cells. The DNA work of Pawliuk et al,18 a high proportion of transduced (10 ␮g per lane) was digested with SacI. This enzyme cuts in both long BM cells carrying the human ferrochelatase cDNA were terminal repeat (LTR) regions in our construction and produces a 3.8 kb selected ex vivo on the basis of the co-expression of the from the LFSN vector. DNA was separated on a 1% agarose gel, trans- + green fluorescent protein (EGFP). The authors demon- ferred to HybondN membrane (Amersham), and hybridized with a ferrochelatase cDNA 32P-radiolabeled probe. 2.5, 5, 25, 50 and 125 pg of strated a complete and long-term correction of photosen- pLFSN were used with 10 ␮g of normal DNA corresponding to 0.25, 0.5, sitivity in this murine EPP model. However, the 2.5, 5 and 12.5 copies per cell of the transgene. The 1.4 kb band corre- approach using vectors encoding gene markers is not sponds to the mouse endogenous ferrochelatase gene which cross- acceptable for use in humans. The aim of our study was hybridized with the human cDNA probe. (b) Ferrochelatase expression in to develop a method of preselecting genetically corrected FDCP1 cells transduced with viral particles from Gp+env86/LXSN or + cells before transfusion using a method based on the Gp env86/LFSN17 clones and selected by G418. expression of the therapeutic gene, rather than on the expression of a potentially toxic marker gene. observed after 24 h of ALA addition (data not shown). Flow cytometry analysis of normal and deficient BM cells is shown in Figure 2a and b, respectively. An analysis Results was also performed with transduced deficient BM cells (Figure 2c) where two distinct peaks appeared. We sorted In vitro transduction of interleukin-3-dependent 35% of the cells showing the lowest fluorescence (Figure hematopoietic precursor cell line FDCP1 2d, open population, denoted LFSN-LowF) and 30% of The Gp+env86/LFSN17 clone was selected for sub- the cells exhibiting the highest fluorescence (Figure 2d, sequent experiments because of its high retroviral pro- closed population, denoted LFSN-HighF). Each of these duction (1.3 × 106 c.f.u.). Supernatants from this clone and two sorted cell populations was then grafted into four the control Gp+env86/LXSN clone were used to trans- deficient recipient mice. duce the FDCP1 cells in Retronectin-coated wells. Two days after the infection, cells were selected in the pres- Chimerism of transplanted animals ence of 1 mg/ml G418 for 2 weeks. A Southern blot The chimerism was similar in the five groups of mice 20 analysis showed that transduced and selected FDCP1 weeks after transplantation: normal mice grafted with cells carried two to three copies of the vector per cell normal BM cells (group 1) 97% (95–99%); deficient mice (Figure 1a). Ferrochelatase activity was 5.4-fold higher in grafted with deficient BM cells (group 2) 98% (95–100%); these cells compared with the LXSN-transduced or non- deficient mice grafted with transduced and nonselected transduced FDCP1 cells (Figure 1b). This clone was there- BM cells (group 3) 99.8% (98–100%); deficient mice fore selected for the transduction of BM cells from EPP grafted with transduced low-fluorescent BM sorted cells mice. (group 4) 98% (95–100%); and deficient mice grafted with transduced high-fluorescent BM sorted cells (group 5) Transduction of BM cells, fluorescence-based selection 98.8% (98–99.5%). and reimplantation into mice Two transduction experiments were performed. The first Hematological, enzymatic and metabolic correction of experiment showed that 25% of colony-forming cells the engrafted mice (CFC) were resistant to G418; 3.5 × 105 of these cells were Hematological parameters (Table 1) observed in mice injected directly into two different EPP mice. The second grafted with transduced and nonselected BM cells (group experiment gave rise to a 33% transduction in CFC. 3) were close to those of EPP mice (group 2). By contrast, Again, 3.5 × 105 of the cells were injected into two mice red blood cell (RBC) count, hematocrit and hemoglobin to complete the group of EPP animals grafted with non- concentration were normalized in mice grafted with selected transduced BM cells. Other 8 × 106 cells (from transduced and low fluorescent-selected BM cells the second experiment) were used for fluorescence-based (group 4). selection of transduced cells. Addition of ␦-aminolevul- Bone marrow cells from animals grafted with trans- inic acid (ALA) induced the intracellular accumulation of duced and nonselected BM cells (group 3) showed a low porphyrins. In preliminary experiments, a clear differ- ferrochelatase activity (Table 2), not significantly different ence of porphyrin accumulation between normal (+/+) from the porphyric mice (group 2). However, cells from and deficient (Fech/Fech) bone marrow cells was animals grafted with low-fluorescent BM cells showed

Gene Therapy Absence of porphyria phenotype following partial gene correction A Fontanellas et al 620

Figure 2 Flow cytometry analysis and sorting of different BM cell populations. Cells were analyzed using UV light excitation (340–360 nm) to characterize porphyrin fluorescence (Ͼ650 nm, channel FL3) into the cells. (a) Nucleated BM cells from a normal BALB/cJ mouse. (b) Nucleated BM cells from a Fechm1pas/Fechm1pas mutant mouse. (c) Results of a 33% transduction of deficient BM cells after 24 h of ALA exposure. The population was then separated into a low-fluorescent (LowF) and high-fluorescent (HighF) fraction. (d) Results of the cell sorting experiment. Open area: cell population corresponding to the low-fluorescent fraction. Closed area: cell population corresponding to the high-fluorescent fraction.

Table 1 Hematological and hepatic parameters of normal and Fechm1Pas/Fechm1Pas mice grafted with normal, deﬁcient and transduced Fechm1Pas/Fechm1Pas bone marrow cells selected or not

Group Graft Recipient RBC Hb Hematocrit Liver/body Alanine Alkaline mice (106 cell/␮l) (g/dl) (%) weight transaminase phosphatase (%) (␮mol/l) (␮mol/l)

1 +/++/+ 8.7 12.1 38.0 4.8 84 118 (n = 6) (7.87–9.53) (11.0–12.9) (34.9–40.1) (4.2–5.0) (82–119) (90–168) 2 −/−−/− 7.75** 9.4** 31.0** 10.1** 731** 1150** (n = 6) (6.5–7.97) (9.3–9.63) (29.6–31.8) (8.5–15.2) (553–956) (689–1348) 3 −/−LFSN −/− 8.0 10.4** 33.8** 11.3** 650** 693**,$ (n = 4) nonselec. (6.69–8.56) (8.71–10.9) (28.1–34.3) (8.5–13.7) (381–1018) (385–1283) 4 −/−LFSN-LowF −/− 10.0$$ 12.8$$ 39.9$$ 7.1*,$$ 666** 690**,$ (n = 4) (8.58–11.2) (11.9–13.2) (35.6–47.5) (3.8–9.2) (47–1150) (62–802) 5 −/−LFSN-HighF −/− 8.0 11.1$ 34.1* 12.7** 929** 799** (n = 4) (7.13–8.47) (10.2–11.7) (31.8–37.3) (10.3–15.2) (510–1046) (711–876)

Group 1, +/+BM cells into +/+mice; group 2, −/−BM cells into −/−mice; group 3, −/−LFSN nonselected BM cells into −/−mice; group 4, −/−LFSN-low ﬂuorescence BM cells into −/−mice; group 5, −/−LFSN-high ﬂuorescence BM cells into −/−mice. All animals were 6 months old at death. *P Ͻ 0.05, **P Ͻ 0.01 versus group 1 and $P Ͻ 0.05, $$P Ͻ 0.01 versus group 2. Data are medians (range).

Table 2 Ferrochelatase activities and protoporphyrin levels of normal and Fechm1Pas/Fechm1Pas mice grafted with normal, deﬁcient and transduced Fechm1Pas/Fechm1Pas bone marrow cells selected or not

Group Graft Recipient Ferrochelatase activity Protoporphyrin level mice (nmol meso-Zn/mg prot.h)

Bone Liver Bone marrow RBC Liver marrow (nmol/g prot.) (␮mol/l) (nmol/g prot.)

1 +/++/+ 1.1 2.78 0.05 0.48 8 (n = 6) (0.54–1.37) (2.0–4.45) (0.03–0.14) (0.23–0.81) (3–41) 2 −/−−/− 0.07** 0.17** 1.65** 5.48** 1256** (n = 6) (0.03–0.113) (0.08–0.19) (0.27–2.09) (4.0–8.21) (430–1966) 3 −/−LFSN −/− 0.16** 0.18** 0.83**,$$ 3.2**,$$ 974** (n = 4) nonselec. (0.01–0.54) (0.03–0.28) (0.27–2.09) (2.4–4.3) (305–1413) 4 −/−LFSN-LowF. −/− 1.31$$ 0.19** 0.18$$ 1.9*,$$ 417** (n = 4) (0.95–2.21) (0.11–0.65) (0.05–0.25) (0.6–2.4) (129–4139) 5 −/−LFSN-HighF −/− 0.24** 0.14** 0.42**,$$ 5.43** 2028** (n = 4) (0.05–0.52) (0.12–0.22) (0.34–0.49) (4.4–7.0) (1165–4156)

Gene Therapy Absence of porphyria phenotype following partial gene correction A Fontanellas et al 621 the same ferrochelatase activity as the normal mice transplantation) or a mild response (the two mice (group 1). engrafted with cells transduced at 33% at the time of BM protoporphyrin levels in EPP mice grafted (Table transplantation; Figure 4b). Three out of four mice 2), with nonselected or low-fluorescent selected BM cells grafted with low fluorescent-selected BM cells and were reduced 1.9- and 9.2-fold, respectively, compared exposed to light did not present any macroscopic dam- with EPP mice, demonstrating the efficiency of the selec- age (Figure 4c), as observed in wild-type mice. Micro- tion of genetically corrected cells. These results were scopically, one mouse presented with only a small confirmed by the study of the percentage of fluorescent localized inflammatory infiltrate in the hypodermis RBC cells shown in Figure 3. This percentage was (Figure 4f). slightly reduced in mice grafted with nonselected BM cells (group 3: 47% versus 61% for group 2, EPP mice) Liver damage but was highly reduced in mice grafted with low fluor- EPP mice showed a marked hepatomegaly when com- escent-selected BM cells (group 4: 19% versus 61% in pared with the control group (10.1% of liver weight per group 2, P Ͻ 0.01). However, correction was not total total body weight versus 4.8%, respectively; see Table 1). when compared with control animals (group 1: 4.7%, P Ͻ 0.01).

Skin photosensitivity To study the cutaneous photosensitivity of the different groups of mice, a dose of 8 J/cm2 was chosen because we observed some apoptotic cells in wild-type mice with the higher dose of 10 J/cm2 (data not shown). Dam- age in skin and the ear extremities began to appear 24 h following light exposure in EPP mice. Five days after irradiation, skin damage in deficient mice was accentu- ated (Figure 4a), whereas we did not observe any lesions on wild-type mice (Figure 4d). Histological sections confirmed these observations with a high percentage of dead cells in the epidermis and dermis, and an extensive inflammatory infiltrate (Figure 4g). Wild-type mice showed no damage microscopically (Figure 4h). Mice grafted with nonselected cells presented either a skin response similar to the EPP mice (the two mice engrafted with 25% of transduced cells at the time of

Figure 4 Macroscopic observations of mice and skin histology 5 days following UVA irradiation. (a) Deficient mice presented with numerous macroscopic skin lesions and yellow ears. (b) EPP mice transplanted with nonselected BM cells presented with variable macroscopic skin lesions. (c) EPP mice grafted with low-fluorescent selected BM cells presented with no macroscopic skin lesion. (d) Wild-type mice presented with no macroscopic skin lesions and transparent ears. Optical microscopic analysis was performed on skin sections stained with hematoxylin–eosin. (e) Non- irradiated deficient mouse skin appeared normal. (f) A mouse grafted with low-fluorescent selected BM cells presented a slight hypodermal (H) inflammatory infiltrate. (g) UVA-irradiated EPP mice showed an epider- Figure 3 Protoporphyrin-mediated fluorescence analysis of RBC. Blood mis stained in red due to eosinophilic cytoplasm of dead cells. This high from the five groups of transplanted mice was diluted in PBS and analyzed number of dead cells induced epidermal disorganization. An inflammatory by flow cytometry as described in Figure 2. Open circles: mice without infiltrate in the dermis was also shown (arrow), combined with hyperplasia any macroscopic or microscopic skin lesions; dotted circle: mouse without of the hypodermis (not shown) inducing a cleavage between the two. (h) macroscopic damage and a small inflammatory infiltrate in hypodermis; Wild-type mice presented neither necrosis or apoptosis in epidermis nor lined circles: mice with low macroscopic lesions and inflamatory infiltrates any dermal inflammatory infiltrate. The apparent difference in skin thick- in epidermis and dermis; closed circles: mice with high cell death rate in ness (e, f, g, h) is due to the angle of section. E: epidermis; D: dermis; H: epidermis and dermis and inflamatory infiltrate. hypodermis. Original magnification ×20. Scale bar: 50 ␮m.

Gene Therapy Absence of porphyria phenotype following partial gene correction A Fontanellas et al 622 No changes in organ weight were observed in mice Discussion grafted with nonselected or high-fluorescent selected cells. By contrast, hepatomegaly was less severe in mice Human EPP is a monogenic disorder characterized clini- grafted with low-fluorescent selected BM cells (7.1% ver- cally by cutaneous photosensitivity, sometimes mild ane- sus 10.1% in EPP mice, P Ͻ 0.01). Normal mice showed mia, and biochemically by the accumulation of excessive typical liver histology with sinusoids carrying blood in amounts of photoreactive protoporphyrin. ␤-Carotene is the parenchymal tissue. Liver from deficient mice effective in preventing skin photosensitivity in some presented large discrete areas delimited by a dense col- cases of EPP.20 Allogeneic BM transplantation might be lagen network, widely called regenerative nodules. The an option to cure this disease. In the unique report of a mean cell area of parenchymal cells in these nodules was bone marrow transplantation in an EPP patient,15 a com- larger (1023 ± 62 ␮m2) than that of parenchymal cells in plete remission of the porphyric phenotype was observed normal mice (498 ± 54 ␮m2, P Ͻ 0.01). Most animals 5 months after transplantation. As in this patient, grafted with low-fluorescent selected BM cells exhibited transplantation of normal BM cells into the reduced regenerative nodules with diffuse limits. Cell Fechm1pas/Fechm1pas mutant mouse was efficient in areas (724 ± 32 ␮m2) showed a tendency to normalization decreasing protoporphyrin accumulation in the blood. It (P Ͻ 0.01 versus EPP mice). can also prevent hepatobiliary complications and reverse Serum from mice grafted either with nonselected or hepatic protoporphyrin accumulation.19 BM cells and low-fluorescent selected transduced BM cells showed specifically hematopoietic stem cells (HSC) are con- reduced alkaline phosphatase levels (Table 1). As sidered as good targets for retrovirus-mediated gene expected, liver ferrochelatase activity was not modified therapy of immunological21,22 and hematological23 genetic after BM transplantation (Table 2). However, the hepatic diseases like EPP. One approach for gene therapy is to protoporphyrin level was reduced in deficient mice introduce ex vivo a normal counterpart of the defective grafted with low-fluorescent selected BM cells (Table 2, gene into the HSC of the affected patient, and to return group 4). it as an autologous BM transplant. To facilitate future ex vivo gene therapy in humans, a critical point is the design Provirus integration in CFC from bone marrow of of efficient selection procedures aimed at increasing the grafted mice frequency of genetically corrected cells before autologous Five months after transplantation, an obvious increase of transplantation. The selection approach uses vectors transduced CFC was observed in mice grafted with low- encoding marker genes, such as GFP,18,24 or genes confer- fluorescent selected BM cells (44%, range 25–52%) when ring cellular resistance to potentially toxic drugs such as compared with animals injected with transduced and neomycin, hygromycin, puromycin and dihydrofolate nonselected BM cells (2.5%, range 0–8%) (Figure 5). reductase (DHFR) or multi-drug resistance (MDR-1) sys- tems.23 A long-term cure of the photosensitivity in the Fechm1pas/Fechm1pas mouse model was obtained by a preselective gene therapy protocol using GFP.18 In the cur- rent study, we wanted to use an approach applicable in humans. Previous work performed on lymphoblastoid cells from two other erythropoietic porphyrias, ie congenital erythropoietic porphyria and hepato-erythropoietic porphyria, demonstrated the feasibility of a fluorescence-based selection of genetically corrected cells.25,26 We applied this system to the BM cells of the EPP mouse model. Because of the specific fluorescence of porphyrins, the development of an efficient strategy of selecting transduced cells depends on a clear difference in porphyrin accumulation between deficient and corrected cells. The presence of exogeneous ALA, a heme precursor, induces an increase in porphyrin fluorescence in deficient cells detected by cytofluorimetry. This selection procedure eliminates the use of a marker gene for determin- ing that the BM cells have received the therapeutic gene. However, the LXSN vector contains the Neo gene, the product of which is an antigenic protein. For human use, a retroviral vector without such a selection gene will have to be considered in view of gene therapy in humans. Animals grafted with transduced and low-fluorescent selected BM cells showed a restoration of BM ferrochelatase activity, in contrast to the low activity remaining in Figure 5 Identification of LFSN proviruses by in vitro amplification the nonselected transduced BM cells. In the same man- (PCR). A total of 25 CFC from BM of each grafted mice was analyzed ner, the porphyrin level was normalized in BM of ani- for the presence of the provirus by nested PCR using specific primers Ј mals grafted with low-fluorescent selected BM cells. designed from a sequence located in the LFSN vector: Fech A: 5 -CCT These results demonstrate the efficacy of the fluor- ATT CAG AAT AAG CTG GC-3Ј, Fech B: 5Ј-CAG AAT AAG CTG GCA CCA TTC-3Ј, Fech C: 5Ј-CTG CCT GTG GTG GAG CAG-3Ј, Fech escence-based selection system. Long-term correction (5 D: 5Ј-AAG CTG CTG CCT GTG GTG-3Ј. The final PCR amplified a months) demonstrated that short- and long-term recon- 270 bp product from human ferrochelatase cDNA. stituting cells were efficiently transduced and selected.

Gene Therapy Absence of porphyria phenotype following partial gene correction A Fontanellas et al 623 Since 44% of CFC carried the provirus and because of Materials and methods a global 120% enzymatic activity in the BM cells, it can be concluded that expression of the transgene was at least Retroviral construct and packaging cell lines two-fold that of the endogenous gene. Strict gene regu- A normal full-length human ferrochelatase cDNA (1.6 kb) lation is not required for EPP, because ferrochelatase is was cloned into the unique EcoRI–BamHI sites of the not the key-limiting enzyme of the heme synthesis path- pLXSN vector31 to obtain pLFSN. The gene of interest is way. Because the porphyric phenotype seemed to be a driven by the MoMLV/MoMSV LTR. The vector contains consequence of the accumulation of protoporphyrin in the NeoR gene which allowed the selection of transduced different tissues, we analyzed this parameter in BM, RBC cells by G418 (Geneticin; Gibco BRL, Grand Island, NY, and liver. Only about 60% of the RBC in the EPP mice USA). The LFSN construct was transfected into the had high protoporphyrin levels. In the erythroid lineage, amphotropic packaging cell line ␺-CRIP by DOTAP protoporphyrin accumulation was highest in erythro- transfection reagent (Boehringer Mannheim, Meylan, blasts and reticulocytes compared with RBC. During the France). Transfectants were selected in the presence of life time of RBC, protoporphyrin can cross the membrane 1 mg/ml G418 for 2 weeks. Filtered supernatants from due to its lipophilic characteristics,27 leading to a virus-producing cells were used to infect ecotropic + ␮ decreased porphyrin concentration with age.28 RBC count Gp env86 cells in the presence of 8 g/ml protamine was normalized in mice grafted with low-fluorescent sulfate (Sigma, St Louis, MO, USA). Two days after trans- transduced BM cells. This means that the erythroid cells duction, cells were selected in the presence of 1 mg/ml of these animals had a longer life due to less direct por- G418 for 2 weeks. Twenty-three resistant clones were iso- phyrin toxicity and less destruction of maturing corrected lated. Viral titers were measured using NIH3T3 cells as a target. Supernatants were tested for the presence of rep- RBC. In this way, Koningsberger et al29 provided evi- lication-competent virus.32 The clones producing the dence that protoporphyrin can induce the generation of highest viral titer were analyzed for ferrochelatase oxygen-free radicals, in particular hydrogen peroxide, expression using the FDCP1 cells. which can yield OH radicals in iron/copper-driven Fen- Packaging and NIH3T3 cell lines were maintained in ton chemistry. The latter compounds are a very potent Dulbecco’s modified Eagle’s medium (DMEM), sup- oxidant; they are able to initiate a chain reaction of lipid plemented with 10% heat-inactivated fetal calf serum peroxidation, to disturb the structure and function of (FCS; BioWhittaker, Emerainville, France), 100 units/ml cellular membranes and to reduce the life RBC. The hem- ° penicillin and 0.1 mg/ml streptomycin at 37 Cin5%CO2 oglobin level was also normalized in animals grafted atmosphere. FDCP1 cells were maintained by standard with selected BM cells. Because the activity of ferrochelat- procedures in ␣-MEM medium, supplemented with 10% 30 ase is rate-limiting for heme synthesis in EPP, restored heat-inactivated FCS, 10% of supernatant of WEHI cells, activity of this enzyme in BM is consistent with an 100 units/ml penicillin and 0.1 mg/ml streptomycin ° increase in heme and hemoglobin in the RBC. at 37 Cin5%CO2 atmosphere. FDCP1 cells were In our experiments, even a partial correction of the BM transduced by retroviral supernatant from the cells led to a therapeutic effect on the EPP mice. Absence Gp+env86/LFSN17 and the Gp+env86/LXSN clones at a of skin photosensitivity was observed in mice with less multiplicity of infection (MOI) of 20 for 24 h. than 20% of fluorescent RBC. Moreover, a partial phenotypic correction was found for animals with 20 to 40% of Preparation and transduction of BM cells from normal fluorescent RBC. and deficient mice Restoration of hepatic function depended on the per- The Fechm1pas/Fechm1pas mutant mice (BALB/cJ) were centage of genetically corrected cells in the graft. Biliary obtained from the Jackson Laboratory (Bar Harbor, ME, excretion was also improved, as indicated by reduction USA). A colony was established in the animal facility of of the alkaline phosphatase level. Reduction of protopor- the Universite´ Victor Segalen Bordeaux 2. All mice were phyrin production from the BM and increased biliary kept under pathogen-free conditions in air-filtered cages excretion resulted in a decrease in protoporphyrin levels and were provided with autoclaved food and water. The in the liver. mice were maintained with unlimited water and stan- In conclusion, fluorescence-based selection of corrected dard laboratory chow under 12 h light/dark cycles. m1pas m1pas cells is an efficient procedure leading to long-term Homozygous males (Fech /Fech ) were crossed m1pas + expression of the transgene from progenitors to differen- with heterozygous females (Fech / ) and homo- tiated hematopoietic cells. In this model, a partial correc- zygous pups were characterized by their high RBC por- tion of bone marrow cells was sufficient to reduce the phyrin concentration. All the animal experimentations were conducted in accordance with the National Institute production of protoporphyrin and its accumulation in of Health Guide for the Care and Use of Laboratory RBC and liver, with an improvement in skin photosensi- Animals. tivity and anemia in EPP mice. The features of this por- Donor male mice (8 to 12 weeks old) were intraperi- phyric mouse mimic those of human erythropoietic pro- toneally injected with 5-fluorouracil (150 mg/kg per toporphyria. These data provide support for the mouse). Five days after injection the donor mice were feasibility of gene therapy in EPP patients with severe killed, then BM cells were collected by flushing the mar- photosensitivity that does not respond to presently avail- row from the two femurs and tibias. Cells were dispersed able methods of photoprotection, and in patients with using a 21-gauge needle and RBC were eliminated by extremely high levels of protoporphyrin that put them hypotonic lysis. Remaining BM cells were washed twice at risk of developing potentially lethal protoporphyrin- with phosphate-buffered saline (PBS) and were incubated induced liver damage. in ␣-MEM medium, supplemented with 12% fetal calf serum (FCS), 12% horse serum, 10% supernatant of

Gene Therapy Absence of porphyria phenotype following partial gene correction A Fontanellas et al 624 WEHI cells, 5% supernatant of COS cells transfected with animals were weighed, then anesthetized with a pento- recombinant murine stem cell factor (rmSCF; Genetic barbital injection and killed by exsanguination. Blood Institute, Cambridge, MA, USA), 100 units/ml penicillin samples were collected through cardiac puncture for ° and 0.1 mg/ml streptomycin at 37 Cin5%CO2 hematological and biochemical parameters. Skin, liver, atmosphere. spleen and BM cells were immediately removed. Liver For the transduction procedure, isolated nucleated BM was weighed before processing samples. Single-cell sus- cells were resuspended at a concentration of 5 × 105 pensions were obtained by passing marrow cells through cells/ml in virus supernatant supplemented with 12% a 21-gauge needle. Red cells were eliminated by hypo- horse serum, 10% of supernatant of WEHI cells and 5% tonic lysis and nucleated BM cell number was estimated of supernatant of COS cells transfected with rmSCF. Cells in a Malassez count slide. Hemoglobin, hematocrit, cell were seeded in RetroNectin (Takara, Boehringer count and volume parameters were measured in a Coul- Ingelheim, Germany)-coated wells and incubated at 37°C ter Counter (Coultronics, Margency, France). Serum, ala-

under 5% CO2 atmosphere for 12 h. Then, the super- nine transaminase and alkaline phosphatase were natant was replaced by fresh medium and incubated for determined by standard methods in a CX7 (Beckman, 36 h. Two days after infection, 2.5 × 104 nucleated BM France). Ferrochelatase activity was measured in the BM cells were seeded in a methylcellulose-based medium and liver according to Camadro and Labbe.34 Protopor- (HCC-4230; StemCell Technologies, Tebu, Meylan, phyrin levels in RBC, plasma and organs were determ- France) with 10% horse serum, 8% supernatant of WEHI ined spectrofluorimetrically35 using a Hitachi F-4500 flu- cells, 4% supernatant of genetically modified COS cells. orescence spectrophotometer (Braun Sciencetec, Les Ulis, Cultures were performed in the presence and absence of France) (excitation 405 nm and emission 605 nm), after G418 to underline CFC at day 12. extraction with 1 mol/l HCLO4/CH3OH (1:1, v/v). Per- centage of fluorescent RBC was scored at the end of the Fluorescence-based selection of retrovirally transduced study. Blood from transplanted mice was diluted in PBS cells and BM transplantation and analyzed by flow cytometry (Coulter Elite) using a Forty-eight hours after the infection of deficient BM cells, 340–360 nm UV light excitation and channel FL3 (Ͼ650 medium was replaced by ␣-MEM medium containing 7% nm) to characterize porphyrin fluorescence. FCS, 1 mmol/l ALA, 2 mmol/l melatonin25 (protecting ␮ agent) and 50 mol/l FeSO4 (Sigma, Saint-Quentin, Detection of provirus integration by polymerase chain France) at pH 7.4. Cell sorting experiments were perfor- reaction (PCR) med 24 h after addition of ALA. Cells were harvested and Fifty thousand BM cells from grafted mice were har- concentrated at 106 cells/ml and sorted by flow cytome- vested and plated in a methylcellulose-based medium as try at a rate of 1500 events/s using an Elite Cytometer described above. CFC were scored on day 12 and a mini- (Coulter, Margency, France). An excitation light source mum of 25 colonies from each mouse was picked, rinsed between 340 nm and 360 nm was used and fluorescence twice with PBS and frozen. Cells were then digested with was measured through a 675 nm band filter. Sorted cells proteinase K in lysis buffer (10 mmol/l Tris-Cl, pH 8.0, were collected in 5 ml sterile tubes containing 0.5 ml FCS. 50 mmol/l KCl, 2.5 mmol/l MgCl , 0.5% Tween-20, 100 Recipient female mice (21–28 days old) were lethally 2 ␮g/ml proteinase K) at 55°C for 2 h, followed by a 10 irradiated with two doses of 6 Gy (␥-rays) spaced 24 h min exposure at 95°C to inactivate the proteinase K. The apart33 to suppress the endogenous hematopoiesis. Each presence of the provirus was characterized by nested recipient received 3.5 × 105 nucleated BM cells from PCR using specific primers designed from a sequence donor male mice 4 h after the second dose of irradiation. located in the LFSN vector: Fech A: 5Ј-CCT ATT CAG Five different conditions of engraftment were tested: AAT AAG CTG GC-3Ј, Fech B: 5Ј-CAG AAT AAG CTG group 1: normal mice grafted with normal BM cells (+/+ GCA CCA TTC-3Ј, Fech C: 5Ј-CTG CCT GTG GTG GAG BM into +/+mice); group 2: deficient mice grafted with CAG-3Ј, Fech D: 5Ј-AAG CTG CTG CCT GTG GTG-3Ј. deficient BM cells (−/−BM into −/−mice); group 3: The final PCR amplified a 270 bp product from human deficient mice grafted with transduced and nonselected ferrochelatase cDNA. Each reaction was analyzed by aga- BM cells (−/−LFSN BM into −/− mice); group 4: deficient rose gel electrophoresis and the specific PCR fragment mice grafted with transduced low-fluorescent BM sorted was detected under a UV light after ethidium bromide cells (−/−LFSN-LowF BM into −/−mice), and group 5: staining. deficient mice grafted with transduced high-fluorescent BM sorted cells (−/−LFSN-HighF BM into −/−mice). Skin photosensitivity and morphological studies Analysis of chimerism in transplanted mice Reversal of skin photosensitivity was assayed 20 weeks The use of males as donors and females as recipients after transplantation using an irradiation protocol. Dorsal made it possible to determine chimerism by fluorescence mouse skin was depilated in order to render mouse skin in situ hybridization (FISH). Bone marrow cells were accessible to irradiation. Twenty-four hours later, mice hybridized with a Y FITC labeled-paint probe for the were irradiated 30 cm from the light source. A dose of 8 mouse chromosome Y and a Cyanin 3 labeled-paint J/cm2 UVA was delivered with a Biotronic device (Vilber probe for the mouse chromosome X (Valbiotech, Paris, Lourmat, Marne La Valleé, France). The UVA lamp deliv- France), as described previously.19 A minimum of 200 ered UV in the range 312–400 nm with a peak at 365 nm. cells was examined. Skin pictures were taken 5 days after irradiation. Sheets of dorsal mouse skin were fixed in 4% formalin, Hematological and biochemical measurements dehydrated, embedded in paraffin, cut in 5 ␮m sections Recipient mice were killed 6 days after UVA exposition and stained with hematoxylin–eosin to visualize the for biochemical, molecular and histological analysis. The irradiation-induced damage.

Gene Therapy Absence of porphyria phenotype following partial gene correction A Fontanellas et al 625 Liver histology 10 Gouya L et al. Inheritance in erythropoietic protoporphyria: a The liver was removed, weighed, rinsed in ice-cold PBS, common wild-type ferrochelatase allelic variant with low divided and fixed in 4% paraformaldehyde by immersion expression accounts for clinical manifestation. Blood 1999; 93: for 48 h. Pieces were then embedded in paraplast, cut in 2105–2110. 5 ␮m slices and the sections were collected on gelatin- 11 Taketani S et al. Structure of the human ferrochelatase gene. 36 Exon/intron gene organization and location of the gene to chro- coated slides and stained with Masson trichrome. The mosome 18. Eur J Biochem 1992; 205: 217–222. sections were then dehydrated, mounted in Eukitt and 12 Wang X et al. Haplotype analysis of families with erythropoietic observed under a Leica LR microscope. Cell surface was protoporphyria and novel mutations of the ferrochelatase gene. estimated from randomly chosen sections in each group. J Invest Dermatol 1999; 113: 87–92. Only cells that had a distinguishable nucleus were ana- 13 Ru¨ fenacht UB et al. Systematic analysis of molecular defects in lyzed. A total of 1000 cells in each group was measured. the ferrochelatase gene from patients with erythropoietic proto- Measurements were performed on an image analysis sys- porphyria. Am J Hum Genet 1998; 62: 1341–1352. tem (Histo200; Biocom, Les Ulis, France). 14 Nordmann Y. Erythropoietic protoporphyria and hepatic complications. J Hepatol Sep 1992; 16: 4–6. Statistical analysis 15 Lichtin A et al. Correction of erythropoietic protoporphyria (EPP) phenotype by allogenic bone marrow transplant. Blood The results are expressed as median and range. One-way 1998; 92 (Suppl. 1): 532a (Abstr. 2146). analysis of variance (ANOVA) was used for comparison 16 Tutois S et al. Erythropoietic protoporphyria in the house of differences between groups of transplanted mice. mouse. A recessive inherited ferrochelatase deficiency with ane- When the group variances were unequal as tested by mia photosensitivity and liver disease. J Clin Invest 1991; 88: Cochran’s test, the data were transformed logarithmically 1730–1736. [log (1 + X)] before analysis. The null hypothesis was 17 Boulechfar S et al. Ferrochelatase structural mutant (Fechm1pas) rejected when P Ͻ 0.05. in the house mouse. Genomics 1993; 16: 645–648. 18 Pawliuk R et al. Long-term cure of the photosensitivity of murine erythropoietic protoporphyria by preselective gene ther- Acknowledgements apy. Nature Med 1999; 7: 768–773. 19 Fontanellas A et al. Reversion of hepatobiliary alterations by We are grateful to M Landry for liver histological analy- bone marrow transplantation in a murine model of erythropo- sis, to F Belloc for flow cytometry analysis, to I Lamrissi- ietic protoporphyria. Hepatology 2000; 32: 73–81. Garcia, S Landriau and M Sanchez for technical assist- 20 Mathews-Roth MM. The treatment of erythropoietic protopor- ance and to JY Daniel and P Costet for the animal facili- phyria. Semin Liver Dis 1998; 18: 425–426. ties. This work was supported by the Institut National de 21 Cavazzana-Calvo M et al. Gene therapy of human severe com- la Sante´ et de la Recherche Me´dicale (INSERM No. 9508), bined immunodeficiency (SCID)-X1 disease. Science 2000; 288: by Association Française contre les Myopathies (AFM), 669–672. by the Spanish Fondo de Investigaciones Sanitarias (FIS 22 Takenaka T et al. Enzymatic and functional correction along No. 00/0446) and Comunidad Autońoma de Madrid with long-term enzyme secretion from transduced bone marrow (CAM No. 08.6/0003/1999.1). The animal facility was hematopoietic stem/progenitor and stromal cells derived from patients with Fabry disease. Exp Hematol 1999; 27: 1149–1159. financed by a grant from the Comite´De´partemental des 23 Nolta A, Kohn D. Haematopoietic stem cells for gene therapy. Pyreńeés Atlantiques de Ligue Nationale contre le Can- In: Potten CS (ed.). Stem Cells. Academic Press: London, 1997, cer, and from Re´gion Aquitaine. A Fontanellas was sup- pp 447–462. ported by the Spanish Instituto de Salud Carlos III (ISCIII 24 Mazurier F et al. Rapid analysis and efficient selection of human No. 98/3165). transduced primitive hematopoietic cells using the humanized S65T green fluorescent protein. Gene Therapy 1998; 5: 556–562. 25 Fontanellas A et al. Fluorescence-based selection of retrovirally References transduced cells in congenital erythropoietic porphyria: direct 1 Kappas A, Sassa S, Galbraith RA, Nordmann Y. The porphyrias. selection based on the expression of the therapeutic gene. J Gene In: Scriver CR, Beaudet AL, Sly WS, Valle D (eds). The Metabolic Med 1999; 1: 322–330. and Molecular Bases of Inherited Disease. McGraw Hill: New York, 26 Fontanellas A et al. Correction of uroporphyrinogen decarboxy- 1995, pp 2103–2160. lase deficiency (hepatoerythropoietic porphyria) in Epstein–Barr 2 Todd DJ. Erythropoietic protoporphyria. Br J Dermatol 1994; 131: virus-transformed B-cell lines by retrovirus-mediated gene 751–766. transfer: fluorescence-based selection of transduced cells. Blood 3 Baart de la Faille H et al. Erythropoietic protoporphyria: clinical 1999; 94: 465–474. aspect with emphasis on the skin. Curr Probl Dermatol 1991; 20: 27 Steinbach P et al. Cellular fluorescence of the endogenous photo- 123–135. sensitizer protoporphyrin IX following exposure to delta-amino- 4 Bloomer JR. The liver in protoporphyria. Hepatology 1988; 8: levulinic acid. Photochem Photobiol 1995; 62: 887–895. 402–407. 28 de Goeij AFPM, Christianse K, van Steveninck J. Decreased 5 Doss MO, Frank M. Hepatobiliary implications and compli- haem synthetase activity in blood cells of patients with erythro- cations in protoporphyria, a 20 year study. Clin Biochem 1989; poietic protoporphyria. Eur J Clin Invest 1975; 5: 397–400. 22: 223–229. 29 Koningsberger JC et al. Exogenous protoporphyrin inhibits Hep 6 Mercurio MG et al. Terminal hepatic failure in erythropoietic G2 cell proliferation increases the intracellular hydrogen per- protoporphyria. J Am Acad Dermatol 1993; 29: 829–833. oxide concentration and causes ultrastructural alterations. J 7 Ishibashi A et al. Erythropoietic protoporphyria with fatal liver Hepatol 1995; 22: 57–65. failure. J Gastroenterol 1999; 34: 405–409. 30 Bloomer JR, Hill HD, Kools AM, Straka JG. Heme synthesis in 8 Lamoril J et al. Human erythropoietic protoporphyria: two point protoporphyria. Curr Probl Dermatol 1991; 20: 135–147. mutations in the ferrochelatase gene. Biochem Biophys Res Com- 31 Miller AD, Rosman GJ. Improved retroviral vectors for gene mun 1991; 181: 594–599. transfer and expression. Biotechniques 1989; 7: 980–982. 9 Sarkany RPE, Alexander GJMA, Cox TM. Recessive inheritance 32 Landau NR, Littman DR. Packaging system for rapid pro- of erythropoietic protoporphyria with liver failure. Lancet 1994; duction of murine leukemia virus vectors with a variable tro- 343: 1394–1396. pism. J Virol 1992; 66: 5110–5113.

Gene Therapy Absence of porphyria phenotype following partial gene correction A Fontanellas et al 626 33 Varas F, Bernad A, Almendral JM, Bueren JA. Relevance of mye- 35 Grandchamp B et al. Studies of porphyrin synthesis in ﬁbro- loablative conditioning in the engraftment of limiting numbers blasts of patients with congenital erythropoietic porphyria and of normal and genetically marked lympho-hematopoietic stem one patient with homozygous coproporphyria. Biochim Biophys cells. Bone Marrow Transplant 1996; 18: 981–989. Acta 1980; 629: 577–586. 34 Camadro JM, Labbe P. A simple ferrochelatase assay. Biochimie 36 Gabe M. Techniques Histologiques. Masson & Co: Paris, 1968. 1981; 63: 463–465.

Gene Therapy