Improved genetic manipulation of human embryonic stem cells Stefan R Braam 1,2 , Chris Denning 3 , Stieneke van den Brink 1 , Peter Kats 4 , Ron Hochstenbach 4 , Robert Passier 1,2 & Christine L Mummery 1,2,5 Low efficiency of transfection limits the ability to genetically manipulate human embryonic stem cells (hESCs), and differences in cell derivation and culture methods require optimization of transfection protocols. We transiently transferred multiple independent hESC lines with different growth requirements to standardized feeder-free culture, and optimized conditions for clonal growth and efficient gene transfer without loss of pluripotency. Stably transfected lines retained differentiation potential, and most lines displayed normal karyotypes. To realize the full potential of hESCs, efficient methods to manip- ulate their genomes are required. hESC lines expressing fluorescent reporters from lineage-specific promoters will be important for selecting specific lineages in which no appropriate cell-surface antigens are expressed and for in vitro toxicological screening. Targeted gene disruption by RNA interference or homologous recombination will facilitate in vitro modeling of human disease when clinically relevant mutations or deletions are known. How- ever, progress has lagged behind expectations in part because of poor transfection and single-cell cloning efficiencies. Lentiviral infection is now the most efficient method for gene transfer but has major limitations, including silencing of randomly integrated copies of the transgene 1 , incompatibility with homologous recom- bination and costly, time-consuming large-scale production. Ade- noviral constructs yield modest (B11%) infection efficiencies 2 , and plasmid transfection shows highly variable transfection effi- ciencies, ranging from 3–35% in independent lines 3 . Furthermore, the most efficient transfection methods have been optimized using two hESC lines, H1 and H9 (WiCell Research Institute). For non- US researchers, or those not funded by the US National Institutes of Health, 4400 other lines are available. Although various stable 3–5 and inducible gene expression systems have been reported for hESCs 6–8 , none have yet been applied to multiple cell lines and growth conditions presently used. In addition, initial gene delivery is often inefficient. Low transfection efficiency in hESC lines therefore remains an unsolved problem. hESCs are usually cultured with mouse embryonic fibroblast (MEF) feeder cells to support self renewal, but more recently, human feeder cells and feeder-free Matrigel (BD Biosciences) substrates have been used in combination with enzymatic or non- enzymatic passage and growth factor–supplemented basal media. To develop a generic method for ectopic gene expression in hESCs, we investigated whether 12 independently derived cell lines (HES-2; Envy; HUES1, 5, 7, 15; HESC-NL1, 2, 3, 4; and NOTT1, 2) could be transferred to common feeder-free culture conditions and undergo efficient transfection using electroporation, lipofection, lentivirus infection and adenovirus infection, without loss of pluripotency or karyotypic stability. The lines we selected had been derived and grown under the most diverse conditions avail- able: mechanical passage on MEFs in serum-containing medium, mechanical passage on human feeder cells in knockout serum replacement (KSR) medium and enzymatic passage on MEFs in KSR medium (Supplementary Methods online). We transferred cells to feeder-free conditions on Matrigel in KSR-containing MEF- conditioned medium 9 in two stages: first, adaptation to Matrigel; second, adaptation to trypsin (if necessary; Fig. 1a). Success, particularly for mechanically passaged lines, was critically depen- dent on very high density culture during the first passages. The combination of these two steps rapidly allowed plating of cells at low density for gene transfer without major loss of cells or pluripotency, as indicated by immunostaining and flow cytometry analyses for cell-surface markers Tra-1-60, GCTM2 and SSEA4, and transcription factors OCT3/4a and SOX2 (Fig. 1b–i, Supplemen- tary Table 1 and Supplementary Fig. 1 online). After adaptation to feeder-free conditions, all hESC cultures were easier to maintain than using their original culture method. We selected 3 cell lines, HES2, HUES7 and HESC-NL4, that were previously maintained in the widest range of conditions, for detailed investigation. We also included 9 other lines in the analysis because the applicability of these methods to multiple lines is of utmost importance. Transfection was efficient in all 12 lines, independent of previous growth on MEF or human feeder cells, enzymatic or mechanical passage or maintenance in serum-replacement contain- ing or serum-replacement conditions (Supplementary Table 1). As transfection of plasmid DNA is of interest for analyses of gene function, we used pCAG-GFP-IRES-Puro r , a plasmid expressing GFP driven by a modified chicken actin promoter, to optimize RECEIVED 4 FEBRUARY; ACCEPTED 13 MARCH; PUBLISHED ONLINE 6 APRIL 2008; DOI:10.1038/NMETH.1200 1 Hubrecht Institute, Developmental Biology and Stem Cell Research, Uppsalalaan 8, 3584 CT Utrecht, The Netherlands. 2 Department of Anatomy and Embryology, Leiden University Medical Centre, Leiden Postal zone S-1-P, P.O. Box 9600, 2300 RC Leiden, The Netherlands. 3 Wolfson Centre for Stem Cells, Tissue Engineering and Modelling (STEM), Centre for Biomolecular Sciences, University of Nottingham, University Park NG7 2RD, UK. 4 University Medical Centre Utrecht, Department of Biomedical Genetics, 3508 AB Utrecht, The Netherlands. 5 Interuniversity Cardiology Institute of the Netherlands, Catherijnesingel 52, 3511, Utrecht, The Netherlands. Correspondence should be addressed to C.L.M. (c.l.mummery@lumc.nl). NATURE METHODS | VOL.5 NO.5 | MAY 2008 | 389 BRIEF COMMUNICATIONS © 2008 Nature Publishing Group http://www.nature.com/naturemethods