Gene Therapy (2005) 12, 852–856 & 2005 Nature Publishing Group All rights reserved 0969-7128/05 $30.00 www.nature.com/gt BRIEF COMMUNICATION Human artificial (HAC) vector provides long-term therapeutic transgene expression in normal human primary fibroblasts

M Kakeda1, M Hiratsuka1,2, K Nagata1, Y Kuroiwa1, M Kakitani1, M Katoh2, M Oshimura2 and K Tomizuka1,2 1Pharmaceutical Research Laboratories, Pharmaceutical Division, Kirin Brewery Co., Ltd, Takasaki-shi, Gunma, Japan; and 2Biomedical Science, Institute of Regenerative Medicine and Bio-function, Graduate School of Medical Science, Tottori University, Nishimachi, Yonago-shi, Tottori, Japan

Human artificial (HACs) segregating freely the generation of cytogenetically normal hPFs harboring the from host chromosomes are potentially useful to ensure both structurally defined and extra HAC vector. This introduced safety and duration of expression in therapeutic gene HAC vector was retained stably in hPFs without translocation delivery. However, low transfer efficiency of intact HACs to of the HAC on host chromosomes. We also achieved the the cells has hampered the studies using normal human long-term production of human erythropoietin for at least 12 primary cells, the major targets for ex vivo gene therapy. To weeks in them. These results revealed the ability of HACs as elucidate the potential of HACs to be vectors for gene novel options to circumvent issues of conventional vectors therapy, we studied the introduction of the HAC vector, which for gene therapy. is reduced in size and devoid of most expressed , into Gene Therapy (2005) 12, 852–856. doi:10.1038/ normal primary human fibroblasts (hPFs) with microcell- sj.gt.3302483; Published online 3 March 2005 mediated chromosome transfer (MMCT). We demonstrated

Keywords: human artificial chromosome (HAC); microcell-mediated chromosome transfer (MMCT); normal primary fibroblasts; transgene expression; erythropoietin (EPO)

Every vector system currently available for gene therapy down), and to assemble a new chromosome de novo has some drawbacks and advantages.1 The use of viral from its constituent DNA elements (bottom-up). Based vectors such as retroviral and lentiviral vectors allow for on our findings that human chromosomes can be utilized the improved duration of therapeutic gene expression as a vector for animal transgenesis,3,4 we have developed by the facile integration of them into the host genome, the methodology to generate top-down HACs by but are plagued by safety concerns such as malignant manipulating human chromosomes with -direc- transformation. On the other hand, nonviral vectors are ted chromosome truncation and homologous recombina- recognized to be nonpathogenic, while less efficient at tion in chicken DT40 cells.5 Using this technology, our introducing and maintaining the transgene expression. group recently constructed a novel human chromosome Artificially engineered human chromosomes are main- 21-derived HAC vector (21DpqHAC vector),6 which was tained independent of the host chromosome through a devoid of most expressed genes by telomere truncation set of cell divisions.2 Accordingly, the host genome is not at the genomic regions proximal to the in disrupted, and the expression of the transgene could be both p (at AL163201 locus) and q (AL163204 locus) arms. sustained for a prolonged period without suffering the A loxP sequence was then introduced into the q arm (at effects of its surrounding sequence on the host genome. AL163203 locus) of the 21DpqHAC vector, which permits These are the ideal properties required for vectors in the site-specific insertion of circular DNA containing the gene therapy, and therefore the utilization of human transgene expression unit by the Cre-loxP system (Figure artificial chromosomes (HACs) could circumvent the 1a). We also showed the microcell-mediated chromo- issues of conventional vectors. some transfer (MMCT)7,8 into an immortalized human There are two basic strategies to construct HAC fibrosarcoma cell line (HT1080) and the mitotic stability vectors:2 to manipulate a natural human chromosome and persistent expression of EGFP gene in the resultant for generating size-reduced minichromosomes (top- HT1080 clones. An autotransplantation of ex vivo transduced normal human primary fibroblasts (hPFs) with the transgene of Correspondence: Dr K Tomizuka, Pharmaceutical Research Laboratories, interest is a potential approach in replacement therapy Pharmaceutical Division, Kirin Brewery Co., Ltd, 3 Miyahara-cho, Takasaki-shi, Gunma 370-1295, Japan for hormone or metabolic enzyme deficiencies, because Received 30 September 2004; accepted 27 December 2004; published the hPFs could be readily obtained from patients and online 3 March 2005 cultured for further genetic modifications. However, the Usefulness of HAC in therapeutic gene delivery M Kakeda et al 853 hPFs carrying the HAC vector, we also showed the long-term expression of therapeutic transgene, human erythropoietin (EPO) as a model, which is a growth factor for erythroid cells and widely used for the clinical treatment of anemia in renal insufficiency.11 Although the introduction of HACs derived from human chromosome 14 fragment (SC20) into normal bovine primary fibroblasts (bPFs) with MMCT was shown for preparing the donor cells of the somatic nuclear transfer to generate trans-chromosomic calves,12 the retention and structure of the HACs in the transduced bPFs after a set of cell divisions remain to be elucidated because of the low transfer efficiency (approximately 1 to 10 Â 10À6 in immortalized cells)7 in MMCT and the limited capacity of bPFs . To investigate the feasibility of minichromosome transfer into normal hPFs by MMCT, we employed HFL-1 cells (RCB0521, RIKEN, Tsukuba, Japan) as recipient hPFs and first examined the SC20 fragment, which contains neor gene and was shown to be stable in various human and mouse cells.13 It was under the condition of 45% polyethylene glycol 1500 containing 10% dimethyl sulfoxide in DMEM that the most frequently G418- resistant colonies were obtained (about 1.26 Â 10À4 of the transduced cells), indicating that the neor marker on the Figure 1 Construction of the EPO-21DpqHAC vector. (a) Schematic SC20 fragment could be transferred into normal hPFs diagram of the construction of the 21DpqHAC vector.6 The distal q arm and this efficiency was better when compared with that was deleted from human chromosome 21 by telomere-directed truncation at in immortalized cells.7 In addition to HFL-1 cells, the AL163204 locus. For the site-directed insertion of the transgene of interest, SC20 fragment was also successfully transferred into a loxP sequence was introduced at AL163203 locus on the q arm. And the other normal hPFs, HUC-F2 cells and HF19 cells then, the distal p arm was further deleted at AL163201 locus as with the q arm deletion. Triangle indicates artificially synthesized telomere sequence (RCB0436, RCB0210, RIKEN, Tsukuba, Japan, data not for telomere truncation. loxP, loxP sequence. (b) Schematic diagram of the shown), suggesting that MMCT of minichromosomes introduction of the epo expression cassette into the loxP site on the could be applicable to a variety of hPFs from patients as 21DpqHAC vector. CMV, CMV promoter; epo, human EPO cDNA, neo, well as above hPFs. neomycin-resistant gene; and bsr, blasticidine-S-resistant gene. pLN1- We then constructed the expression cassette of human EPO was constructed as follows: SV40 polyA sequence was amplified with epo gene, which is under the control of the CMV primer pairs, 50-CGGGATCCCTCGAGCGAGACATGATAAGATACA TTGATG-30 and 50-GGAAGATCTTCCTAATCAGCCATACCACATTT promoter and SV40 polyA signal, as an application GTAGAGG-30, and pSTneoB25 as a template DNA, using KOD-PLUS example, and introduced them into the 21DpqHAC DNA polymerase (Toyobo, Osaka, Japan). This DNA fragment was vector by Cre-loxP recombination in CHO (#21)–hyg4 digested with BamHI and BglII, and then cloned into the BamHI site of cells6 (EPO-21DpqHAC vector, Figure 1b). We obtained pBS226 (Gibco BRL, Grand Island, NY) as for pBS226-pA. CMV 21 G418-resistant clones (named H4E cells), and all of promoter sequence was amplified with primer pairs, 50-CGGAATTCC 0 0 which secreted EPO into conditioned media (250– GGACATTGATTATTGACTAGTTATTAATAG-3 and 5 -CGGGATCCC 6 GGGTGTCTTCTATGGAGGTCAAAACAG-30, and pBS226 as a tem- 520 IU/10 cells/6 days). The retention of markers for plate DNA, and then cloned into EcoRI–BamHI site of pBS226-pA as for human chromosome 21 (D21S275, PRED65, PRED3) on pLN1. For pLN1-EPO, human EPO coding sequence was amplified with the EPO-21DpqHAC vector was confirmed by PCR. primer pairs, 50-CGGGATCCCGGCCACCATGGGGGTGCACGAAT To examine the structure and copy number of the GTC-30 and 50-CGCTCGAGCGCTATCTGTCCCCTGTCCTGCAGG-30, EPO-21DpqHAC vector in H4E cells, fluorescence in situ and human epo cDNA as a template DNA, and then cloned into BamHI– hybridization (FISH) analysis was carried out on four XhoI site of pLN1. Arrows with short-dashed line indicate XbaI digestion site for Southern blot analysis. Arrows indicate the primer positions for clones using human Cot1 DNA as the probe. In all four M13RV, SVNp1 and NeoRp2. clones, a single spot signal was observed in most of the metaphase specimens (93–99%, n ¼ 50 on each clone), indicating that the EPO-21DpqHAC vector was retained ability of 21DpqHAC as a vector for ex vivo therapeutic as a single copy per cell. No insertion or translocation or gene delivery in human primary cells has not been obvious size alteration of the EPO-21DpqHAC vector evaluated in the above study. Here, we studied the was observed cytogenetically. Then, we used one of introduction of the 21DpqHAC vector into normal hPFs H4E clones (H4Ec10) to transfer the resultant EPO- with MMCT,7,8 which is recognized as the most suitable 21DpqHAC vector into HFL-1 cells with MMCT and method for the transfer of a single, intact chromosome obtained 16 colonies from five MMCT experiments under from one cell to another and enabled the lower incidence blasticidine-S (bsd) selection. of fragmentation and rearrangement of donor chromo- To assess the retention of the EPO-21DpqHAC vector somes compared to transfection-based methods.9,10 To in HFL-1 cells, the presence of the neo gene and the epo our knowledge, this is the first report to demonstrate that expression cassette was determined by PCR on 11 of 16 the structurally defined, extra artificial chromosome is bsd-resistant clones (Figure 2a). In addition, these 11 highly maintained in normal hPFs without accompany- clones produced EPO into conditioned media (260– ing aberration of host chromosomes. In the resultant 690 mIU/106 cells/24 h), indicating that the EPO-

Gene Therapy Usefulness of HAC in therapeutic gene delivery M Kakeda et al 854 2b, Table 1). This clearly demonstrated the introduction of the EPO-21DpqHAC vector into normal hPFs. There is no evidence of any insertion or translocation of the vector sequence on host chromosomes or obvious size alteration of the EPO-21DpqHAC vector itself. Further- more, no aberration of the epo expression cassette was confirmed by Southern blot analysis (Figure 3). These results indicate that the karyotype of HFL-1/EPO clones was apparently normal and the overall structure of EPO- 21DpqHAC vector was not changed in these cells as with the observation in the HAC donor H4Ec10 cells. The EPO-21DpqHAC vector was retained at a constant copy number mostly in a single copy per cell (Table 1), whereas in the case of the human X centromere-based minichromosomes their copy number per cell was increased when the overall size of the minichromosome was substantially reduced.15 To investigate the mitotic stability of the EPO-21DpqHAC vector in normal hPFs, HFL/EPO clones were cultured and maintained in the absence of the selection drug, and then FISH analysis was carried out probed with 21aDNA. The retention rates of the EPO-21DpqHAC vector were initially 88.2 to 100% of metaphase, and after six to nine passages (around 10 to 14 PDL, Table 1), they were 65.5 to 90.9%, indicating that the EPO-21DpqHAC vector is stable in normal hPFs under nonselective condition. Mitotic-loss rates were 0.94 to 2.3% per generation, comparable with those of the 21DpqHAC vector in human fibrosarcoma cell line HT1080 cells (o0.42%).6 Figure 2 Introduction of the EPO-21DpqHAC vector into normal hPFs. This level of stability is also comparable with those MMCT was carried out as follows: HFL-1 cells were cultured in T-25 flask reported for the other top-down (0.22–1.05%; Mills et al,15 (5 Â 105 cells) in DMEM containing 20% FBS (Gibico BRL) at 371C, 40.15%; Saffery et al16) and bottom-up (0.1–0.5%; 6.5% CO2, and washed with DMEM. Microcells purified from H4E cells 17 18 7 Harrington et al, 0.48–1.6%; Ikeno et al, 0.49–1.7%; (6 Â 10 cells) were overlaid on HFL-1 cells in DMEM containing 50 mg/ 19 ml Phytohemagglutinin-P and incubated at 371C for 15 min. DMEM was Mejia et al ) artificial chromosomes in HT1080 cells. then removed and 2 ml of 45% PEG1500–10% DMSO solution was added HFL-1/EPO clones proliferated from a single colony up (PEG1500, Roche, Indianapolis, IN, USA; DMSO, Sigma, St Louis, MO, to 3.8 Â 107 cells in six to nine passages, and then their USA). After 1 min at room temperature, 45% PEG1500–10% DMSO growth stopped, suggesting that cell senescence occurred solution was removed, and HFL-1 cells were washed with DMEM and and the introduction of the EPO-21DpqHAC vector did cultured in DMEM containing 20% FBS for overnight. Cells were then not cause immortalization. replated onto five 48-well plates and selected under blasticidine-S (3 mg/ml) D for 2 weeks. (a) PCR analysis for genomic DNA of HFL-1/EPO cells. To evaluate how long the EPO-21 pqHAC vector Genome, genomic DNA of recipient HFL-1 cells, D2S1334, STS marker allows the expression of human epo transgene, four HFL- for human chromosome 2, neo, neomycin-resistant gene; and epo cassette, 1/EPO clones were cultured at a confluent monolayer, the expression cassette of human epo gene. Genomic were subjected and the amount of EPO secreted in conditioned media to PCR using Ex Taq DNA polymerase (TaKaRa-bio, Otsu, Japan). Primer was sequentially determined (Figure 4). For at least 12 0 0 0 pairs were as follows: 5 -AGGCCTCTCTTCGAATGATT-3 and 5 - weeks, the EPO expressions were sustained without loss CTCCATGATCCAGTCACCTC-30 for D2S1334, M13RV (50-CAGGA 0 0 0 or attenuation, and clone #1-1 provided the maximum AACAGCTATGAC-3 ) and NeoRp2 (5 -CCTGCAGTTCATTCAGGG-3 ) 6 for neo gene, SVNp1 (50-CGGGATCCCTCGAGCGAGACATGATAA EPO expression up to 18 IU/10 cells/24 h, demonstrat- GATACATTGATG-30) and NeoRp2 for epo. (b) A typical metaphase ing that the EPO-21DpqHAC vector is able to provide FISH image of HFL-1/EPO clone #1-1. Arrowhead indicates the EPO- prolonged expression of the transgene in normal hPFs. 21DpqHAC vector. Hybridization was carried out as described,4 probed The EPO expression gradually increased, correlating with the human alphoid DNA sequence derived from the centromere region with the time course. Although the mechanism remains a 14 of chromosome 21 (21 DNA). to be elucidated, it is known that transgene products are synthesized at increased rates in a confluent monolayer and under cell growth-arrested conditions.20 During the preparation of this paper, the transfer of 21DpqHAC vector was successfully introduced into other bottom-up artificial chromosomes (ACEs) contain- HFL-1 cells (HFL-1/EPO cells). To examine the structure ing RFP gene into human primary mesenchymal stem of transferred EPO-21DpqHAC vector in HFL-1/EPO cells (MSCs) by lipofection was reported.21 Although the clones, FISH analysis was carried out on four clones RFP expression and retention of ACEs were shown in a probed with the human alphoid DNA sequence derived small population (7–15%) of the transduced MSCs up to from the centromeric region of human chromosome 21 2–3 weeks, the long-term expression and mitotic stability (21aDNA),14 which is also hybridized with that of of ACEs were not examined. In addition, the structure of chromosome 13. In all of the four clones, one or two RFP-ACEs and the copy number of introduced RFP gene extra spot signals were observed independent of the four were inconclusive. On the other hand, the structure of signals on the centromere of host chromosomes (Figure our 21DpqHAC vector is highly predictable from DNA

Gene Therapy Usefulness of HAC in therapeutic gene delivery M Kakeda et al 855 Table 1 Mitotic stability of the EPO-21DpqHAC in normal hPFs

Clone Passage Number of metaphase Number of 21aDNA signals/cell Retention rate of HAC (%) Loss rate (R)

4+0 4+1 4+2

#1-1 Initial 38 3 35 0 92.1 #1-7 Initial 50 0 28 22 100.0 #3-1 Initial 57 3 54 0 94.7 #5-7 Initial 34 4 30 0 88.2

#1-1 9 29 10 19 0 65.5 0.023 #1-7 6 44 4 22 18 90.9 0.0094 #3-1 6 40 9 31 0 77.5 0.019 #5-7 6 37 11 26 0 70.2 0.022

21aDNA14; a human alphoid DNA probe from the centromeric region of chromosome 21. Mitotic-loss rates of the EPO-21DpqHAC vector n were calculated by the following formula: Nn ¼ N0 Â (1ÀR) , where N is the retention rate after ‘n’ generations, N0 is the retention rate at day 0 and R is the loss rate per generation. Passages 6 and 9 were equivalent to 10 and 14 generations, respectively.

Figure 4 Long-term EPO expression in conditioned media (CM) of HFL- 1/EPO clones under nonselective condition. HFL-1/EPO clones were cultured in collagen-I-coated six-well culture plates (BD Biosciences, San Jose, CA, USA) at full confluence. Media were replaced and after 24 h, and the CMs were harvested. The concentration of human EPO in CM was determined by Quantikine IVD Human EPO Immunoassay kit (R&D Systems, Minneapolis, MN, USA).

epo gene by a retroviral vector were encapsulated within subcutaneously implanted immunoisolation devices (0.81–6.6 Â 106 cells/kg), and EPO was secreted into the circulation showing a higher mean hematocrit than that of the control.23 In this study, HFL-1/EPO clones could be expanded up to 3.8 Â 107 cells, which is comparable to the baboon model. Although the expression level of EPO in HFL-1/EPO clones (o18 IU/106 cells/24 h) was lower than that in the above bMSC system (183–712 IU/ 106 bMSCs/24 h), improved expression should be Figure 3 Southern blot analysis for the epo expression cassette of HFL-1/ achieved by using the promoters and regulatory ele- EPO clones. Genomic DNAs were digested with XbaI (Roche, Indiana- ments better suited for gene expression in normal hPFs. polis, IN, USA) and applied to 0.8% agarose gel electrophoresis. Blotting Thus, the MMCT is applicable to prepare large numbers 3 32 and hybridization was carried out as described using P-labeled human of normal hPFs expressing therapeutic transgenes on the epo cDNA probe. Arrow indicates the expected band size for the epo expression cassette. HACs. One of the important issues for gene therapy vector is to establish physiological levels of transgene expression sequences of human chromosome 21.22 The site-directed in specific cells or tissues for the appropriate period in a introduction of the epo expression cassette is also defined normally regulated manner. Further optimization of by the single loxP sequence,6 which is supported in regulatory elements, such as hypoxia-responsive enhan- Figure 3. cer, which can be activated through the transcriptional In a previous baboon ex vivo EPO-replacement model, complex hypoxia-inducible factor 1A,24 might permit the baboon primary MSCs (bMSCs) transduced with human physiological control of EPO expression in normal hPFs.

Gene Therapy Usefulness of HAC in therapeutic gene delivery M Kakeda et al 856 Our HAC vector could carry large-sized DNA,5,12 and (X/autosome translocation) via micro-cell fusion. Jpn J Cancer therefore the utilization of genomic region including Res 1989; 80: 122–125. regulatory sequences and the gene of interest itself, such 8 Killary AM, Fournier RE. Microcell fusion. Methods Enzymol as insulin for diabetes and large genes for monogenic 1995; 254: 133–152. disorders, might establish physiological control of 9 Klobutcher LA, Miller CL, Ruddle FH. Chromosome-mediated transgene expression. gene transfer results in two classes of unstable transformants. In conclusion, this study shows the potential applica- Proc Natl Acad Sci USA 1980; 77: 3610–3614. tion of the 21DpqHAC vector in hormone-replacement 10 de Jong G et al. Efficient in-vitro transfer of a 60-Mb mammalian gene therapy, which might ensure safety with no lesion artificial chromosome into murine and hamster cells using of host chromosomes and provides long-term therapeu- cationic lipids and dendrimers. Chromosome Res 2001; 9: 475–485. tic transgene expression. This also implies that utilization 11 Goodnough LT, Monk TG, Andriole GL. Erythropoietin therapy. N Engl J Med 1997; 336: 933–938. of the HAC vector offers novel options for ex vivo gene 12 Kuroiwa Y et al. Cloned transchromosomic calves producing therapy. human immunoglobulin. Nat Biotechnol 2002; 20: 889–894. 13 Shinohara T et al. Stability of transferred human chromosome fragments in cultured cells and in mice. Chromosome Res 2000; 8: Acknowledgements 713–725. We thank Drs I Ishida, T Mikayama and T Inoue-Nishida 14 Ikeno M, Masumoto H, Okazaki T. Distribution of CENP-B for comments and discussions. We also thank Ms M boxes reflected in CREST centromere antigenic sites on long- range alpha-satellite DNA arrays of human chromosome Takahashi and Mr F Ayabe for technical assistance. This 21. Hum Mol Genet 1994; 3: 1245–1257. study was partly supported by a Health and Labor 15 Mills W, Critcher R, Lee C, Farr CJ. Generation of an Sciences Research Grant for Research into Human approximately 2.4 Mb human X centromere-based minichromo- Genome and Tissue Engineering from the Ministry of some by targeted telomere-associated chromosome fragmenta- Health, Labor and Welfare in Japan (MO, KT), and by tion in DT40. Hum Mol Genet 1999; 8: 751–761. the 21st Century COE program from the Ministry of 16 Saffery R et al. Construction of neocentromere-based human Education, Culture, Sports, Science and Technology of minichromosomes by telomere-associated chromosomal trunca- Japan (MO, KT). tion. Proc Natl Acad Sci USA 2001; 98: 5705–5710. 17 Harrington JJ et al. Formation of de novo and construction of first-generation human artificial microchromo- References somes. Nat Genet 1997; 15: 345–355. 18 Ikeno M et al. Construction of YAC-based mammalian artificial 1 Thomas CE, Ehrhardt A, Kay MA. Progress and problems with chromosomes. Nat Biotechnol 1998; 16: 431–439. the use of viral vectors for gene therapy. Nat Rev Genet 2003; 4: 19 Mejia JE et al. Functional complementation of a genetic 346–358. deficiency with human artificial chromosomes. Am J Hum Genet 2 Larin Z, Mejia JE. Advances in human artificial chromosome 2001; 69: 315–326. technology. Trends Genet 2002; 18: 313–319. 20 Fussenegger M et al. Controlled proliferation by multigene 3 Tomizuka K et al. Functional expression and germline transmis- metabolic engineering enhances the productivity of Chinese sion of a human chromosome fragment in chimaeric mice. Nat hamster ovary cells. Nat Biotechnol 1998; 16: 468–472. Genet 1997; 16: 133–143. 21 Vanderbyl S et al. Transfer and stable transgene expression of a 4 Tomizuka K et al. Double trans-chromosomic mice: maintenance mammalian artificial chromosome into bone marrow-derived of two individual human chromosome fragments containing Ig human mesenchymal stem cells. Stem Cells 2004; 22: 324–333. heavy and kappa loci and expression of fully human antibodies. 22 Lander ES et al. Initial sequencing and analysis of the human Proc Natl Acad Sci USA 2000; 97: 722–727. genome. Nature 2001; 409: 860–921. 5 Kuroiwa Y et al. Manipulation of human minichromosomes to 23 Bartholomew A et al. Baboon mesenchymal stem cells can be carry greater than megabase-sized chromosome inserts. Nat genetically modified to secrete human erythropoietin in vivo. Biotechnol 2000; 18: 1086–1090. Hum Gene Ther 2001; 12: 1527–1541. 6 Katoh M et al. Construction of a novel human artificial 24 Ebert BL, Bunn HF. Regulation of the erythropoietin gene. Blood chromosome vector for gene delivery. Biochem Biophys Res 1999; 94: 1864–1877. Commun 2004; 321: 280–290. 25 Katoh K, Takahashi Y, Hayashi S, Kondoh H. Improved 7 Koi M, Morita H, Shimizu M, Oshimura M. Construction of mammalian vectors for high expression of G418 resistance. Cell mouse A9 clones containing a single human chromosome Struct Funct 1987; 12: 575–580.

Gene Therapy