Targeting of the Human Coagulation Factor IX Gene at rDNA Locus of Human Embryonic Stem Cells

Background Genetic modification is a prerequisite to realizing the full potential of human embryonic stem cells (hESCs) in human genetic research and regenerative medicine. Unfortunately, the random integration methods that have been the primary techniques used keep creating problems, and the primary alternative method, gene targeting, has been effective in manipulating mouse embryonic stem cells (mESCs) but poorly in hESCs. Methodology/Principal Findings Human ribosomal DNA (rDNA) repeats are clustered on the short arm of acrocentric chromosomes. They consist of approximately 400 copies of the 45S pre-RNA (rRNA) gene per haploid. In the present study, we targeted a physiological gene, human coagulation factor IX, into the rDNA locus of hESCs via homologous recombination. The relative gene targeting efficiency (>50%) and homologous recombination frequency (>10−5) were more than 10-fold higher than those of loci targeted in previous reports. Meanwhile, the targeted clones retained both a normal karyotype and the main characteristics of ES cells. The transgene was found to be stably and ectopically expressed in targeted hESCs. Conclusion/Significance This is the first targeting of a human physiological gene at a defined locus on the hESC genome. Our findings indicate that the rDNA locus may serve as an ideal harbor for transgenes in hESCs.


Introduction
Over the past thirty years, the combined use of murine embryonic stem cells (mESCs) and gene targeting, which allows researchers to study gene function in vivo, has revolutionized developmental research [1]. Parallel to mESCs, human embryonic stem cells (hESCs) can also proliferate indefinitely and differentiate into multiple lineages both in vitro and in vivo [2]. More importantly, because they have a human genetic background, they may be powerful tools in the study of human genes and in regenerative medicine. Unfortunately, the gene targeting strategy that is most widely used to manipulate mESCs has worked poorly in hESCs due to their resistance to non-viral transfection and sensitivity to single-cell cloning. So far, only a dozen sites have been successfully targeted in hESCs [3][4][5][6][7][8][9][10][11][12][13][14][15]. Random transgene integration is the method of genetic modification most commonly used with hESCs; but the uncertainty of the integration site leads to other problems. A transgene may become silenced if it is integrated into a heterochromatin area, or it may disrupt or activate endogenous genes, leading to apoptosis or otherwise changing the cell's fate. In addition, because it is thought that random integration is mediated by non-homologous end joining (NHEJ), an imprecise DNA-repair mechanism, integration may be incomplete and plasmid debris may end up integrated into the genome [16,17]. These problems may be solved if the transgene is targeted to an appropriate site where it can be expressed without any serious functional consequences [18].
Human 45S ribosomal DNA (rDNA) is clustered on the short arm of all five acrocentric chromosomes (chromosomes 13, 14, 15, 21, and 22). It consists of approximate 400 copies of the 45S pre-RNA (rRNA) gene per haploid [19]. The rRNA gene is transcriptionally active, producing approximately 80% of the total RNA in rapidly dividing cells. Loss or gain of the short arm of acrocentric chromosomes is common in humans and does not usually have any phenotypic effects and can be inherited stably. Balanced translocations between the short arm of an acrocentric chromosome and the other chromosomes are also observed without apparent phenotypic effect. For this reason, we presumed that transgenes targeted into the rDNA locus would be transcriptionally active without any unexpected effects. The rDNA locus is a candidate harbor favorable for effective and safe transgene cell manipulations. A recent report revealed that the rDNA cluster exhibited strikingly variable lengths between and within human individuals and showed high intrinsic recombinational instability during both meiosis and mitosis [20]. This indicates that a high rate of gene targeting may be achieved. The transgene targeted into the rDNA locus is expected to work in hESCs.
In this report, we targeted human coagulation factor IX, a physiological gene, into the rDNA locus via homologous recombination and found the process to be highly efficient. We here demonstrated that the integrated clones retained the main characteristics of ES cells. After long-term culture, the targeted cells retained a normal karyotype and expressed the transgene stably. Our findings provide a new strategy for manipulating hESCs in both basic and applied research.

Results
We constructed an rDNA-targeting plasmid, pHrnF9, which introduced a promoterless neomycin resistance (neo) cassette and an EF1-a driven human coagulation factor IX (F9) open reading frame into the 45S pre-RNA gene. The two cassettes were flanked by a 59 long homologous arm (4.5 kb) and a 39 short homologous arm (1.1 kb). The first cassette contained an encephalomyocarditis virus internal ribosomal entry site (EMCV-IRES), which enabled resistant gene expression under the control of endogenous RNA polymerase I (Pol I) promoter upstream after homologous recombination ( Figure 1).
Prior to targeting hESCs, we tested the new plasmid construct in HT1080 cells via electroporation. We did two gene targeting experiments on HT1080 cells (Table S1). In one of the experiments, three million HT1080 cells were exposed to a single 2000 V, 50 mF pulse with 20 mg linearized pHrnF9 at room temperature using the BioRad Gene Pulser II (0.4 cm gap cuvette, BioRad, Hercules, CA, U.S.). Up to 400 mg/ml of G418 (Sigma, St Louis, MO, U.S.) was added 72 hours after electroporation. Finally, 997 resistant clones were obtained (Table S1). Out of the 39 clones screened by PCR, 12 were identified as homologous recombinants. This indicated a relative targeting efficiency of 31% and an absolute targeting frequency of 0.01%. These values are higher than those of any previous gene targeting experiment performed on HT1080 cells. The targeted clones secreted F9 protein stably (Table S2).
For gene targeting of hESCs, we nucleofected trypsinized H9 hESCs, using 10 mM rho-associated kinase (ROCK) inhibitor Y-27632 two hours before and during the 24 hours immediately after nucleofection to improve single-cell survival. This method yielded sufficiently high transfection efficiency (Figure 2a). G418 selection was initiated 48-72 hours after transfection, starting at low drug concentrations and slowly building up to 50 mg/ml. Because H9 cells are highly sensitive to G418, the initial drug concentration of 25 mg/ml was sufficient to kill untransfected cells. Drug-resistant clones were picked and expanded for genotyping two weeks after nucleofection. In a typical experiment, 3.2610 6 cells were nucleofected with 5 mg linearized pHrnF9, and 57 drug-resistant clones were obtained. Of the 22 clones analyzed, 14 targeted clones were identified by PCR ( Figure 2d) and Southern blot analysis (Figures 2f and 2g). Relative targeting efficiency was 64% (14/22) and the absolute homologous recombination frequency was 1.13610 25 . Both of these values are higher than those reported in previous studies other than studies of artificial zinc finger nucleases (ZFNs). Generally, dozens of clones were obtained for each case of single nucleofection followed by two weeks of selection. In a series of experiments, similar results were obtained regularly (Table 1).
Following successful gene targeting of hESCs, we analyzed Gbanded chromosomes on nine targeted clones after genotyping, and all showed normal karyotypes (Figure 2e). To confirm that the targeted clones retained hESC characteristics after gene targeting, we examined hESC marker expression. Immunocytochemistry and alkali phosphatase (AP) staining revealed that the targeted hESCs lines expressed stage-specific embryonic antigen (SSEA)-3 ( Figure 3d Figure 3c). To determine the differentiation potential of the targeted clones, we cultured the cells in suspension to form embryoid bodies (EBs). After 7 days, the EBs were transferred to gelatin-coated plates and cultured for another 7 days. Immuno- To evaluate pluripotency in vivo, we injected the targeted cells subcutaneously into immuno-compromised mice. Two months later, the targeted cells generated various complex teratomas comprising structures and tissues derived from the three embryonic germ layers, including gallbladder (endoderm) (Figure 5a (Figure 5f). Our data indicate that the rDNAtargeted clones retained their ES characteristics.
To determine whether targeted transgene could be expressed at the rDNA locus, reverse transcription PCR (RT-PCR) analysis was carried out, revealing that wild-type H9 did not express endogenous F9 and that the targeted clones did, even after 40 passages of contiguous culture (Figure 6a). Quantitative analysis using enzyme linked immunosorbent assay (ELISA) showed that the targeted clones expressed the transgene at different levels ( Figure 6c), indicating the site-specifically integrated transgene could be expressed in the targeted clones. It also showed that the synthesized protein could be secreted into the supernatant. Western blot analysis showed that the F9 protein could be detected in both the cell lysate and concentrated supernatant ( Figure 6b and Figure S1). The different levels of transgene expression observed among different targeted clones may have been caused by the different targeted rDNA copies, which had different transcriptional levels.

Discussion
In recent years, hESCs have been studied eagerly by thousands of laboratories, but results have been disappointing because of their low rate of transgenesis and homologous recombination relative to mESCs. One of main characteristics of hESCs is the maintenance of normal karyotypes after long-term culture. To achieve this, they must maintain some kind of self-repairing mechanism (such as homologous recombination) through hundreds of mitotic divisions. For this reason, we think the intrinsic frequency of homologous recombination may not be as low as reported and the real barrier to hESC gene targeting is their resistance to clonal expansion. This problem can be solved by using of neurotrophin cocktail or ROCK inhibitor [21,22]. For instance, once treated with Y27632, the survival rate of single hESCs can be improved by more than two orders of magnitude without sacrificing any ES properties [22]. By this means, we achieved homologous recombination rate of more than 10 25 in hESCs. In other studies, the difficulty of gene targeting has differed from site to site [11]. The addition of Y27632 can only improve clonal expansion, which directly results in increased numbers of drug-resistant clones; but this does not, in theory, markedly alter the ratio of homologous recombinants to non-homologous recombinants. Normally the targeting efficiency of both mESCs and hESCs is about 1%, but in our experiments, it was more than 50%. The gene-trap strategy that we used can partially explain this, especially considering that this strategy has been reported to be more effective than positive-negative selection [3,23]. However, the main reason would seem to be the relatively high intrinsic activity of homologous recombination at the rDNA locus.
Although there are about 400 copies of rDNA repeats per haploid, it is, to our knowledge, uncertain whether the high frequency of homologous recombination is due to the copy number of target sites. A previous study proved that targeting does not depend on the number of targets in mammalian cells [24]. However, recent study using ZFNs did obtain higher rate of gene targeting in cells with two target sites than in cells with only one [25].
ZFNs have been widely used to achieve efficient homologous recombination of gene targeting vectors with various endogenous loci in cultured and primary mammalian cells [26], even those with very short homologous arms [27]. The customized artificial nucleases can introduce DNA double strand breaks at target sites and then stimulate the cell's endogenous homologous recombination machinery. In this system, a donor DNA can replace any lost portion of its corresponding chromosomal segment at high efficiency. In 2009, two research groups successfully performed gene targeting in hESCs using ZFNs. Hockemeyer et al. obtained 40 homologous recombinants among 42 resistant clones (relative targeting efficiency 94%) using their most efficient ZFN pair [28].An absolute targeting frequency of 0.14-0.24% was achieved by Zou et al. [10]. ZFNs are usually superior when used to generate a mixed population of targeted and untargeted cells without drug selection. However, the use of ZFNs can be compromised by their  laborious design process, their toxicity, and other undesired effects.
In gene targeting of hESCs with the purpose of generating sitespecific edited recombinants, our strategy described here is more effective and safer than the ZFNs method. Human ROSA26 and ENVY loci are transcriptionally active and considered to be candidate sites for targeted transgene integration [5,29]. But it is uncertain whether the cells will exhibit functional consequences when one of the two copies is disrupted. Another hESC line for gene targeting has been reported. In this line, a LoxP-docking site was introduced in to the HPRT locus [12]. This cell line can be used for targeting exogenous sequences using the Cre-Lox system. Because the first round of targeting is favorable to an active X chromosome, disruption of the HPRT locus may lead to a Lesch-Nyhan phenotype. In contrast, the human rRNA cluster consists of hundreds copies of tandemly repeated rDNA units. Variations in rDNA copy number are common among healthy individuals and balanced chromosomal translocation involving the rRNA cluster occurs without any apparent phenotypic effect. These properties indicate that the rDNA locus may hold a high intrinsic homologous recombination activity and facilitate effective transgene expression. In the present study, the rDNA locus was targeted with high efficiency, no undesired effect was detected, and the transgene was expressed stably in the integrated clones.
It has been reported that RNA polymerase II (Pol II)transcribed genes can be silenced at the rDNA locus in yeast cells, especially under an active Pol I promoter [30,31]. In the   current study, we screened the resistant clones using a promotertrap strategy. This means that the neo gene could only be expressed under a transcribed Pol I promoter. As a result, the clones cannot be selected unless the transgene is targeted into an active rDNA repeat in a dynamic chromatin structure. However, we found that all the five selected clones expressed the Pol II gene. This differs from the results of a previous study conducted in yeast. The question raised by our results is whether the mechanism of the Pol II gene silencing at the rDNA locus is similar in all eukaryotes, from yeast to human cells. Restructuring of the rDNA clusters has been observed among somatic cells [20]. This raises concerns about the potential instability of any inserted transgene in the rDNA locus and the risk that the transgene might be translocated among the rDNA clusters or lost during mitotic expansion. However, the odds of this happening to one specific targeted copy out of the hundreds of rDNA copies in the human genome are very low. They are rendered even lower by the fact that the restructuring of the rDNA clusters during mitosis is rare.
In summary, using the strategy described here, we integrated human coagulation factor IX into the rDNA locus of hESCs via homologous recombination. Compared to other sites reported, this locus proved to be subject to efficient targeting. The targeting of hESCs at the rDNA locus did not change the main ES characteristics of the cells and the transgene was expressed stably in targeted hESCs. This is the first gene targeting of a human physiological gene at a defined locus on hESCs. Our findings indicate that the rDNA locus may serve as an ideal harbor for transgenes in hESCs.

Construction of Plasmid and Gene Targeting
For gene targeting, one passage prior to transfection, hESCs were detached by 1 mg/ml dispase and replated on dishes coated with Matrigel TM (BD Biosciences, San Jose, CA, U.S.) for three days in MEF conditioned hESC medium. Two hours before transfection, 10 mM Y27632 was added to the medium. For targeting, the hESCs were incubated at 37uC with trypLE TM Select (Invitrogen, Carlsbad, CA, U.S.) for 3 minutes. Then the cells were immediately collected and counted. The centrifuged cells were resuspended with 100 mL Human Stem Cell Nucleofector Kit 2 (Lonza, Basel, Switzerland) and 5 mg linearized pHrnF9 and nucleofected using Nucleofector II (Lonza, Basel, Switzerland) using program A023. After recovery in 500 mL RPMI-1640 (Hyclone, Beijing, China) for 5 minutes at room temperature, the transfected cells were plated on PMEF-NL (Millipore, Bedford, MA, U.S.) in hESC medium containing 10 mM Y27632 (Sigma, St Louis, MO, U.S.). G418 selection was initiated 48-72 hours after transfection, depending on the cell density. The final concentration of G418 was 50 mg/ml. About two weeks after transfection, a portion of resistant clones was picked and the remaining clones were fixed and stained with Giemsa stain. Clones with diameters of $2 mm were considered resistant.

PCR and RT-PCR
Genomic DNA was isolated from the cells using phenol/ chloroform extraction. The primers used are as follows: P1 59-GGG TGG GGC AGG ACA GCA AGG GGG AGG AT -39; P2 59-GGC GAT TGA TCG GCA AGC GAC GCT CAG ACA G -39.

Alkaline Phosphatase Staining and Immunofluorescence
Three day old cell clumps were fixed with 4% paraformaldehyde. After washing in TBST, alkaline phosphatase was stained with 0.375 mg/ml nitrobluetetrazolium chloride and 0.188 mg/ ml 5-bromo-4-chloro-3-indolyl-phosphate (Roche, Indianapolis, IN, U.S.), pH 9.5, for 15 minutes in darkness for detection of alkaline phosphatase. For immunofluorescent staining, cells on gelatin-coated coverslips were fixed in 4% paraformaldehyde and permeabilized with 0.1% Triton-X 100. After blocking with 10% normal donkey serum (Jackson ImmunoResearch, West Grove, PA, U.S.), the cells were incubated with the first antibody against

In vitro Differentiation
Three day old cell clumps were incubated with 1 mg/ml dispase for 10 minutes at 37uC then washed with DMEM/F12. After culture on ultra-low attachment plates for 7 days in hESC medium without bFGF, EBs were transferred to gelatin-coated coverslips and cultured for another 7days. Differentiated cells were analyzed by immunofluorescence. Directed differentiation was performed according to the protocol of the National Stem Cell Bank of America (https://www. wicell.org/index.php?option=com_docman&task=doc_download &gid=1064).

Teratomaformation and Analysis
Targeted cells from one 10 cm dish were dissociated with 0.05% trypsin/EDTA (Invitrogen) and collected in DMEM/F12. After being washed with DMEM/F12, cells were resuspended in 140 mL DMEM/F12 and 70 mL Matrigel TM (BD Bioscience, San Jose, CA, U.S.). The suspension was injected subcutaneously into the hind legs of immunocompromised mice. Eight to twelve weeks later, the formed teratomas were removed and fixed in 4% paraformaldehyde overnight. The fixed tissues were sectioned and stained with hematoxylin and eosin. All procedures regarding the care and use of animals are in accordance with institutional guidelines. This study was approved by the Ethics Committee of State Key Laboratory of Medical Genetics of China, No. 2008-ANIMAL-004.

Western Blot
Protein samples were electrophoresed and electrotransferred onto PVDF membranes (Millipore, Bedford, MA, U.S.). Blots were incubated with a primary antibody anti-human Factor IX (Affinity Biologicals, Canada) overnight at 4uC. The blots were than incubated with horseradish peroxidase-conjugated secondary antibodies(Sigma, St Louis, MO, U.S.) for 1 hour at room temperature and detected with an ECL system (Amersham Biosciences, Piscataway, NJ, U.S.). Prestained molecular weight standards (Fermentas, Glen Burnie, MA, U.S.) were used to estimate the apparent molecular weight.

ELISA
After culturing in hESC medium for three days, 24-hour-old supernatants were collected from six-well plates. Total cells (together with MEFs) and MEFs from parallel wells (subtracting the number of MEFs from the total number of cells) were trypsinized and counted. All supernatants were collected in triplicate. ELISA was performed using Paired Antibodies for ELISA-Factor IX (Cedarlane Laboratories, Ltd., Burlington, Canada) according to the manufacturer's instructions. Reference curves were constructed using serial dilutions of normal pooled plasma (Pacific Hemostasis, Cape Town, South Africa), with correlation coefficient (R 2 ) of at least 0.990 using a 5-parameter logistic curve fit algorithm.