Therapy (1999) 6, 931–938  1999 Stockton Press All rights reserved 0969-7128/99 $12.00 http://www.stockton-press.co.uk/gt expression, but not , is improved by adhesion-assisted lipofection of hematopoietic cells

H Keller1, C Yunxu2, G Marit2, M Pla3, J Reiffers2, J The`ze1 and P Froussard1 1Unite´ d’Immunoge´ne´tique Cellulaire, De´partement d’Immunologie, Institut Pasteur, Paris; 2Laboratoire de Greffe de Moelle, URA CNRS 5540, Universite´ Victor Segalen Bordeaux 2, Bordeaux; and 3Immunoge´ne´tique de la souris, U93 INSERM, Hoˆpital Saint- Louis, Paris France

In contrast to adherent cells, cells growing in suspension in up to 30% of transfected human primary T lymphocytes. and particularly hematopoietic cells, are notoriously difficult Flow cytometry analysis performed on T lymphocyte sub- to transfect in vitro using nonviral approaches. In the sets revealed that 8 and 9%, respectively, of CD4 and CD8 present study, the effect of cell adhesion on gene transfer cells could be transfected with a carrying the green efficacy was investigated by allowing hematopoietic cells fluorescent gene. Other adherent cells, such as to bind to an adherent cell monolayer (ACM) before being MS5 murine stromal cells or HeLa epithelial cells, were subjected to cationic liposome-mediated DNA transfer. also a compatible matrix for AAL. Moreover, the pCMV␤ Human CD34 and T CD4 cell lines were cultivated on an plasmid was present in similar amounts in the nuclei of TF1 ACM constituted of murine fibroblast NIH3T3 cells and cells transfected in suspension or with the AAL procedure. transfected with a plasmid carrying the ␤-galactosidase These data raise the possibility that cell matrix/ gene. X-gal staining showed that up to 27% of the cells hematopoietic cell interactions might govern expression of expressed the transgene. In contrast, less than 0.1% of the transgene in hematopoietic cells growing usually in these cells were positively transfected in suspension. This suspension, but not endocytosis of liposome/DNA particles adhesion-assisted lipofection (AAL) procedure was also and plasmid migration to the . successfully tested on blood lymphocytes, since it resulted

Keywords: liposomes; cationic liposomes; adhesion-assisted lipofection; gene transfer; hematopoietic cells

Introduction obtained using adherent cells lines.5,6 However, many cell types, particularly primary cells, blood lymphocytes and At present, methods using viral vectors are the most more generally hematopoietic cells growing in suspen- promising for delivery and expression of foreign sion are largely resistant to this form of gene transfer,7 in vitro and in vivo. However, safety concerns and the though successful has been reported in difficulty in obtaining large quantities of recombinant mouse bone marrow cells.8 viral vectors have prompted the search for efficient, non- The sequence of cellular and molecular events underly- immunogenic and easy-to-prepare nonviral vector sys- ing the complex phenomenon of DNA transfer remains tems. Among them, cationic liposome-mediated gene highly speculative,9 but it is commonly accepted that 1–3 delivery, which was pioneered by Felgner, has been endocytosis is the main mechanism of entry of demonstrated to be an effective approach presenting a liposome/DNA particles,10–13 and relatively high trans- 4 low risk of intrinsic toxicity and pathogenicity. Cationic fection efficiency is due to the intrinsic membrane-rup- liposomes, which are easy to prepare on a large scale, turing capability of cationic liposomes as a result of form complexes with DNA through charge interactions destabilizing the plasma and/or endosome membrane.14–16 in such a way that DNA is protected from degradative In contrast to adherent cells, liposome/DNA particles activities and virtually any size of plasmid molecule can have been reported to bind inefficiently to the cell surface be packaged. Liposome/DNA complexes bind to the of lymphocytes, suggesting that particles attach to mem- negatively charged cell surface due to the presence of brane components involved in cell anchoring to the extra- excess positive charges in the complex; the nature of non- cellular matrix and composed of cadherins or proteogly- specific interactions results in the ability to drive cans present on adherent cells only.6 This weak expression of a foreign gene in host cells, including non- interaction between particles and membrane lympho- dividing cells, with the greatest transfection efficiency cytes would prevent endocytosis of liposome/DNA par- ticles and release of DNA into the cytosol of nonadherent cells. Because the ability of target cultured cells to be Correspondence: P Froussard, Unite´ d’Immunoge´ne´tique Cellulaire, De´partement d’Immunologie, Institut Pasteur, 75724 Paris cedex 15, transfected with the liposome/DNA complex relies France essentially on their adhesion capacity, we investigated Received 8 May 1998; accepted 21 December 1998 the role of cell adhesion in gene delivery to hematopoietic Adhesion-assisted lipofection in hematopoietic cells H Keller et al 932 cells. We primarily observed that hematopoietic cells could be transfected when cultured on an adherent cell matrix before being subjected to liposome-mediated DNA delivery. We then studied the usefulness of this new gene transfer approach in more detail using several types of established hematopoietic cell lines and primary lymphocytes of human origin, as well as different cell matrices. The role of cell adhesion in gene transfer optim- ization was also investigated.

Results

Coculture with adherent cells makes TF1 cells competent for gene transfer Figure 2 Analysis of ␤-galactosidase transcripts in cells transfected with ␤ Human TF1 cells, which grow in suspension, were liposome/pCMV complexes. Cells were transfected using the AAL pro- cedure or a standard lipofection (SdL) assay. Total RNA was purified from loaded on to an NIH3T3 ACM as described in Materials 5 × 105 transfected cells, subjected to DNase I and amplified by RT-PCR. and methods. The whole mixed cell population was then PCR products encoding ␤-galactosidase DNA sequences were then elec- subjected to liposome-mediated DNA transfer using trophoresed, transferred to nitrocellulose and hybridized with a 32P-labeled /pCMV ␤ complexes. Two days after trans- specific DNA probe. from untransfected cells and RNAs subjected fection, adherent cells were eliminated by adhesion on to both DNase 1 and RNase A did not show any RT-PCR products. plastic flasks and expression of the bacterial ␤-galacto- sidase gene was assessed in purified TF1 cells and in pure by the anti-mouse mAb. In other respects, respectively TF1 cells transfected in suspension. Counting of dark 97.1% and 91.9% of purified transfected cells were blue-stained cells revealed that up to 26% of TF1 purified detected with anti-HLA class I and anti-CD34 mAbs, cells expressed ␤-galactosidase activity transferred with compared with 99.5% and 94.6% for pure TF1 cells. The the pCMV␤ vector. In contrast, less than 0.1% of blue same transfection study performed using MS5 stromal cells was observed when TF1 cells were transfected in cells as a matrix showed that virtually 100% of recovered suspension (Figure 1). Total RNA was purified from cells suspension cells were recognized with both anti-HLA harvested 48 h after transfection and ␤-galactosidase class I and anti-CD34 mAbs (Figure 3b). Taken together, transcripts were subjected to a non-quantitative RT-PCR. these representative experiments show that NIH3T3 and Bacterial ␤-galactosidase mRNA level in TF1 cells trans- MS5 contaminant cells are barely detectable in purified fected with AAL was 50 to 80% of what was found in suspension cells and that essentially pure TF1 cells can NIH3T3 cells; transgene transcripts were poorly detected be recovered after the three 2-h adhesion steps. in TF1 cells transfected in suspension (Figure 2). Thus, it is obvious that culture of TF1 cells on a NIH3T3 mono- AAL, but not adhesion, is optimized by prolonged layer authorizes a substantial improvement in liposome- coculture of TF1 cells with the cell matrix mediated gene transfer. In the interest of optimizing the AAL procedure, TF1 cells and NIH3T3 ACM were cocultured for increasing per- Purified TF1 cells transfected with AAL are essentially iods of time before being subjected to transfection. Pur- free of contaminating adherent cells ified TF1 cells were then assessed for bacterial ␤-galacto- The purpose of these experiments was to assess the sidase . As shown in Figure 4, the efficiency of the three 2-h adhesion steps used at the end percentage of TF1 blue cells increased linearly for the first of the AAL procedure. In practice, the purity of the cell 6 to 8 h of coculture before reaching a plateau, but acqui- suspension was evaluated by detection of specific cell sition of the full competent state by TF1 cells required surface markers. As shown in Figure 3a, less than 1.1% overnight coculture with the ACM before transfection. of the purified AAL transfected cells could be recognized The ability of TF1 cells to bind to the ACM was studied as a function of the time to link the gene transfer efficiency to the adhesion/suspension state of TF1 cells. Surprisingly, 72% of TF1 cells were already bound to the cell matrix less than 2 h after being loaded on to the ACM. This indicates that prolonged coculture with the ACM is necessary for optimized liposome-mediated gene delivery and that adhesion of TF1 cells to the ACM pre- cedes the acquisition of the features rendering TF1 cells competent.

CD34 and CD4 cell lines are efficiently transfected with AAL As a step towards examining primary cells, we examined ␤ the ability of the AAL procedure to mediate gene deliv- Figure 1 Analysis of -galactosidase expression in purified TF1 cells after ery in three different CD34 cell lines (TF1, KG1a, K562 transfection with lipofectamine/pCMV␤ complexes. TF1 cells were trans- fected in suspension (a) or using the adhesion-assisted lipofection pro- cells) and in Jurkat CD4 T lymphocytes using NIH3T3, cedure on to a NIH3T3 cell matrix (b). Forty-eight hours after gene deliv- MS5 or HeLa ACMs; all suspension cells bound rapidly ery, the purified TF1 cells were stained using X-gal. to the different ACMs. Two days after transfection, pur- Adhesion-assisted lipofection in hematopoietic cells H Keller et al 933

Figure 3 Purity of TF1 cells collected after the three 2-hour-adhesion steps. Purified CD34 TF1 cells transfected with adhesion-assisted lipofection performed with NIH3T3 (left) or MS5 (right) adherent matrix were examined by one-color flow cytometry. Murine MHC-H-2d antigen expressed at the surface of NIH3T3 cells was detected with an antimouse mAb to quantify the NIH3T3-contaminating cells. Human TF1 cells were identified both Ј with anti-HLA class 1 and anti-CD34 mAbs. The second staining was achieved with FITC or R-PE-labeled goat antimouse F(ab )2 fragment. The number shown in each quadrant indicates the corresponding percentage of cells with respect to the total number of analyzed cells. Displayed on the x axis is the fluorescence intensity on a logarithmic scale and on the y axis is the relative cell number. Pure NIH3T3 and TF1 transfected cells were used as a control.

(Figure 5). Lipofection of Jurkat cells on NIH3T3 ACM resulted in 18.5% transfection efficiency, as compared with 0.8% in controls, but the best experiments permitted ␤-galactosidase expression in up to 27% of the cells. AAL performed with the MS5 matrix gave essentially the same results as with NIH3T3 ACM, though CD4 cells were sometimes observed to be more permissive to DNA transfer with the MS5 matrix. Thus, up to 31% of Jurkat cells were transfected on the MS5 ACM. Use of HeLa matrix did not give such good data since only 4.9% and 3.5% of TF1 and KG1a cells turned blue. This nevertheless corresponded to strong enhancement of gene transfer efficacy compared with transfection of suspension cells; but the possibility that some blue cells counted as TF1 cells might be HeLa cells, cannot be rejected since these cells were more difficult to eliminate. Altogether, these results demonstrate that AAL permits an efficient trans- Figure 4 Gene transfer efficacy in TF1 cells and adhesion of TF1 cells to fection into a broad range of hematopoietic cell lines, and the cell matrix as a function of time of suspension/adherent cell coculture. that many adherent cells can be used as a matrix. TF1 cells and NIH3T3 ACM were cocultured for increasing periods of time before being subjected to transfection. Purified TF1 cells were then assessed for bacterial ␤-galactosidase gene expression, as described in PBL subjected to AAL are permissive to gene transfer Materials and methods. The percentage of TF1 blue cells was plotted as We next examined the performance of AAL in human a function of time of coculture before transfection. Adhesion of TF1 cells to the NIH3T3 ACM was also studied as a function of the time of cell PBL. Forty hours after transfection, PBL represented less coculture. than 15% of the whole mixed cell population, probably because primary lymphocytes were slowly or non-divid- ing cells. X-gal staining of purified lymphocytes revealed ified cells were examined for ␤-galactosidase activity. that up to 30% and 12% of PBL transfected on NIH3T3 Pure CD34 cells and CD4 lymphocytes were also trans- or MS5 ACM expressed the transgene, in contrast to less fected in suspension. As expected, transfection perfor- than 0.1% for cells transfected in suspension (Figure 5). med with NIH3T3 ACM raised the percentage of blue- Thus, the substantial enhancement of efficiency of stained cells from less than 0.1% in the standard protocol DNA transfer into human PBL with the NIH3T3 and MS5 up to 26%, 16% and 19% in TF1, KG1a and K562 cells cell matrices was comparable to that observed above in Adhesion-assisted lipofection in hematopoietic cells H Keller et al 934

Figure 5 Gene transfer to several lineages of human hematopoietic cells and to human primary lymphocytes. TF1, KG-1a, K562, Jurkat cells and human PBL were transfected with AAL using liposome/pCMV␤ com- Figure 6 EGFP expression in CD4 and CD8 lymphocyte subsets. PBL ␤ plexes and several cell matrices. Cells were transfected either in suspension were transfected on a NIH3T3 monolayer with plasmid pCMV or or on to a NIH3T3, MS-5 or HeLa cell monolayer. Successful gene trans- pEGFP. The mixed NIH3T3/PBL cell population was analyzed by flow fer was measured by X-gal staining as outlined in Materials and methods. cytometry two days after transfection for EGFP expression in CD4 and Single asterisk and double asterisks indicate assays showing gene transfer CD8 lymphocyte subsets. The gates corresponding to CD4 or CD8 nega- ␤ efficiency of less than 0.1% or 0.03%, respectively. Transfection of K562, tive cells have been cleared from the Figures. Plasmid pCMV was used Jurkat and PBL on HeLa cell matrix cells was not done (nd). to distinguish positive cells expressing EGFP from negative cells and back- ground fluorescence.

hematopoietic cell lines, although primary lymphocytes to a Southern evaluation procedure enabling quantifi- were non-dividing cells. cation of the ␤-galactosidase gene. A study performed 48 h after transfection revealed that EGFP is expressed in primary CD8 and CD4 purified nuclei of TF1 cells transfected either in suspen- lymphocyte subsets sion or on a cell matrix contained a comparable copy The purpose of this work was to evaluate the perform- number of plasmid molecules, ie about 200 plasmid mol- ance of AAL in primary T lymphocyte subsets, directly ecules per nucleus, while 200 to 350 plasmid copies were in the whole mixed target/matrix cell population without found in the nucleus of NIH3T3 cells (Figure 7). One the requirement for the three 2-h adhesion step. As week after transfection, a mean of 8 pCMV␤ plasmid described above, PBL were loaded on an NIH3T3 ACM molecules was detected in nuclei of transfected cells before being transfected with liposome/pEGFP com- while transfection efficiency was considerably affected, 5 plexes. At day 2, cells were incubated with anti-CD4 or to 10 times less than at day 2, in terms of number of cells anti-CD8 mAbs. Cell labeling with R-PE-labeled anti- expressing the transgene; the rapid decrease of the trans- mouse Fab fragment enabled us to thoroughly dis- gene in transfected cells is consistent with what has been tinguish CD4 and CD8 lymphocyte subsets from murine reported previously for transient transfection. Most of the cells. The mixed NIH3T3-PBL cell population was then plasmid molecules were found in the nucleus rather than analyzed by two-colour flow cytometry for EGFP in lym- in the cytosolic compartment of transfected cells. Because phocytes carrying either the CD4 or the CD8 cell surface it is well-known that cell lysis with NP-40 leads to nuclei marker. Scoring for cells expressing the EGFP protein without an outer membrane, it can be hypothesized that revealed that 8% and 9%, respectively, of CD4 and CD8 detected pCMV␤ DNA molecules were localized inside lymphocytes were fluorescent cells (Figure 6). Controls the nucleus rather than bound to the outer nuclear mem- performed on cells transfected in suspension showed no brane of transfected cells. It can be concluded from these significant EGFP fluorescence. data that liposome particles containing plasmid DNA molecules enter the TF1 cells by endocytosis and are then Plasmid molecules enter the cells whether transfected in conveyed to the nucleus whether these cells are culti- suspension or with AAL vated in suspension or on a cell monolayer. A search for pCMV␤ DNA sequences was undertaken in an attempt to follow the migration of plasmid molecules Discussion through the cell membrane and to the nucleus of trans- fected TF1 cells. Great care was taken in the isolation of In this report, we have demonstrated a transfection clean and unruptured nuclei free from residual plasmid method termed adhesion-assisted lipofection which leads not contributing to gene expression and which might be to a substantial improvement in liposome-mediated gene trapped in the cell membrane or cytoplasm. Total plas- delivery in vitro into human hematopoietic cells. AAL mid DNA contained in transfected cell nuclei, whether achieved transfection with high efficiency not only of dif- integrated into cellular DNA or not, were then subjected ferent cultured cell lineages such as CD34 cells and CD4 Adhesion-assisted lipofection in hematopoietic cells H Keller et al 935 tion but on the time of coculture before transfection. Moveover, purified TF1 cells transfected with AAL and overexpressing the ␤-gal protein were shown to contain similar amounts of the transgene transcript to those found in matrix cells, but it was barely detected in TF1 cells transfected in suspension. Finally, direct FACS analysis of TF1-NIH3T3 mixed cell population demon- strated that at least 14% of TF1-cells were recognized as expressing the EGFP protein when cells were transfected using the AAL procedure and pEGFP plasmid (data not shown). Our results shed light on the possible implications and consequences of cell-to-cell adhesion on liposome/DNA complex routing and transgene expression. Our model, using the same cell either in a suspension or in an adhesion state allowed the question of the mechanism of liposome-mediated DNA transfer into adherent and non- adherent cells to be addressed. Transfection of several ␤ hematopoietic cell types growing in suspension was Figure 7 Southern blot analysis of pCMV DNA in transfected cell demonstrated here to be enhanced at least 200-fold when nuclei. Cells were transfected using the AAL procedure or a standard (SdL) lipofection assay. At days 2 and 7 after transfection, total nucleic these cells were cultured on to an adherent cell mono- acids were extracted from 5 × 105 purified transfected cell nuclei and layer. AAL efficiency performed with the MS5 matrix digested by EcoRV and NotI. For Southern analysis, one tenth of digestion was comparable to that with NIH3T3 ACM for all types products were electrophoresed in 1.0% agarose gel, transferred to nitrocel- of hematopoietic cells tested. HeLa ACM increased gene 32 ␤ lulose and hybridized with a P-labeled -galactosidase DNA probe. The delivery, but with less efficacy. Preliminary experiments hybridization signal corresponding to the 1.2 kb DNA fragment was performed with human endothelial cell matrix showed a quantified using a Phosphorimager and compared with serial dilutions of digested pCMV␤ DNA. The evaluated number of pCMV␤ plasmid mol- transfer efficiency of about 3% both with TF1 and Jurkat ecules per nucleus of transfected cells was noted below the corresponding cells. Thus, murine fibroblast cells and to a lesser extent, lane of every transfection assay (means ± s.d. of four or five human epithelial-like and endothelial cells, shared fea- determinations). tures rendering human suspension cells permissive to liposome-mediated DNA transfer. On the other hand, it is probable that gene delivery into the different cell lines lymphocytes, but also of PBL, primary CD4 and CD8 and into primary cells was regulated by a common path- lymphocyte subsets. This indicates that the AAL pro- way since all cell types were transfected with the same cedure is not restricted to established cell lines, but can range of efficiency on a particular ACM. In other respects, also be utilized for primary cells. Moreover, we recently studies investigating the role of stroma or extracellular established that AAL was able to drive ␤-galactosidase matrix in retroviral infection revealed that the presence of gene expression in up to 27% of primary enriched hema- a cell feeder layer during protocols increases topoietic stem cells from human cord blood (data not gene transfer efficiency.17–20 This enhanced transduction shown). Thus, AAL works independently of the target is probably due to the colocalization of and cells, since all cell types were transfected with the same target cells on specific domains of fibronectin molecules range of efficiency, and is effective both in dividing and allowing hematopoietic cell transduction by exploiting nondividing cells. In addition, this transfection procedure unique ligand–receptor interactions.17,21,22 In this system, does not require a prior stimulation of the cells with a interactions between cells and fibronectin molecules were chemical agent. mediated via cell surface proteoglycan molecules, VLA- Because of the unusual features of the cell matrix in 4 and/or VLA-5 integrins expressed at the surface of improving the gene transfer, the possibility of a passive hematopoietic cells. The possibility that, in AAL, fib- diffusion of the ␤-gal protein from the cell matrix to the ronectin or other cell-adhesion molecules carrying human hematopoietic cells was explored. As shown in anionic residues and expressed at the surface of adherent Figure 1, positive hematopoietic cells were strongly cells only, might bind both hematopoietic cells and cat- stained which corresponds to large amounts of protein ionic liposomes, thus authorizing their colocalization and in the cells. In contrast, passive diffusion of an exogenous endocytosis of particles into the suspension cells, is not protein into the TF1 cells would have produced weakly supported by our data. Indeed, the use of fibronectin- stained cells. Figure 4 shows assays in which TF1 cells coated flasks failed to enhance the number of TF1 cells were cocultivated 1 h to 14 h with NIH3T3 cells before expressing the transgene significantly (data not shown). transfection; in all these assays, most of the TF1 cells were Additional experiments showed us that growth of TF1 attached to the cell matrix after a 2-h coculture and TF1 cells in NIH3T3 culture medium supernatant, as well as cells were intimately imbricated with NIH3T3 cells 2 days culture of TF1 cells on anti-CD34 -coated flasks, after transfection. Thus, in all assays, TF1 cells were essentially failed to improve gene transfer (data not grown and transfected in the same cell matrix environ- shown). In addition, the term ‘transfected in suspension’ ment and would have been comparably contaminated by is inaccurate because all TF1 cells attach strongly to the released ␤-galactosidase in the event of protein diffusion plastic when grown in serum-free medium and in the from the cell matrix to hematopoietic cells. This was not presence of liposome/DNA complexes. On the other the case because the number of blue TF1 cells did not hand, similar amounts of plasmid molecules were found depend on their coculture with the matrix after transfec- to be conveyed to the nucleus of TF1 cells transfected in Adhesion-assisted lipofection in hematopoietic cells H Keller et al 936 suspension or cocultured on a cell matrix and to the HeLa are adherent cells. Growth factor dependent TF1 nucleus of NIH3T3 transfected cells. Altogether, these cells were maintained with 10 ng/ml GM-CSF. Peripheral data show that adhesion of hematopoietic cells on culture blood mononuclear cells (PBMC) were prepared from flasks or a cell monolayer does not favor entry of plasmid heparinized venous blood of healthy adult volunteers. into the cells. This also demonstrates that the plasma Depletion of monocytes by two 30-min adhesion steps on membrane, which has been described as the first critical a plastic flask resulted in less than 2% of the peripheral barrier that must be penetrated in order to achieve highly blood lymphocytes (PBL) expressing the monocytic CD14 efficient gene delivery into the cytoplasm of cells1,14,15 marker at their surface. does not specifically prevent liposome/DNA complex delivery into hematopoietic cells, but probably acts as in Liposome adherent cells. Lipofectamine reagent (GIBCO BRL, Eragny, France) This article reports that adhesion of suspension cells to used in these studies is a polycationic lipid composed of the ACM appeared clearly to precede, but not to a positively charged lipid, DOSPA, and a neutral lipid, accompany, the acquisition of the features allowing an DOPE, in a 3:1 molar ratio. optimized expression of the transgene. These results and evidence that plasmid molecules are equally conveyed to Plasmid the cell nucleus whether the cells are cultivated in sus- The pCMV␤ plasmid (Clontech, Palo Alto, CA, USA) car- pension or on a cell matrix support the hypothesis that rying a full-length E. coli ␤-galactosidase gene under con- suspension-cell-to-adherent-cell adhesion might be neces- trol of the cytomegalovirus (CMV) immediate–early pro- sary but that further cell interactions with the ACM after moter and enhancer was used to assess transgene adhesion are required for an efficient expression of the expression in most of our studies. pEGFP-N3 (Clontech) transgene; these cell interactions might govern the trans- carrying the enhance green fluorescent protein (EGFP) gene expression by stimulation of cellular factors acting was used specifically for study of gene delivery into at the transcriptional level. The fact that DNA uptake and whole PBL and T-lymphocyte subsets. were transgene expression do not always correlate in trans- purified with the Qiagen EndoFree Plasmid Kit (Coger, fected cells, but that processes subsequent to DNA Paris, France). uptake could be even more important in determining gene expression, has been widely documented.16,23 Thus, delivery of plasmid DNA to cultured epithelial cells was Adhesion-assisted lipofection (AAL) 5 demonstrated not to be a limiting factor, since virtually Adherent cells (10 ) were plated in 1 ml of complete Dul- 100% of cells receiving a lipid–DNA conjugate rapidly becco medium per well of a 24-well dish (TTP, ATGC transport exogenous DNA to the region of the cell Biotechnologie, Noisy le Grand, France) and grown for nucleus but only a small portion of these cells, are able 6 h to approximately 70% confluency to form the adher- × 5 to express the DNA at detectable levels.24,25 We con- ent cell monolayer (ACM). At that time, 1.25 10 cells × 5 firmed here that delivery of plasmid DNA into cultured (TF1, K562, KG1a, Jurkat cells) or 2.0 10 primary lym- cells would not be the only limiting factor, but would phocytes in 1 ml RPMI medium were loaded on to the implicate features inherent in the cells as contributing to ACM; unless otherwise stated, the mixed cells were the poor efficiency of lipid-based gene expression cocultivated overnight to allow the suspension cells to observed in vitro. The putative cellular factors involved adhere to the ACM. According to the manufacturer’s pro- ␮ ␤ ␮ in expression of a transgene transferred with AAL could tocol, 0.5 g of plasmid pCMV in 25 l of serum-free ␮ be compared with the regulators expressed by estab- OptiMEM medium was gently mixed with 4 g of lipo- ␮ lished stromal cells, which control the maintenance of fectamine in 25 l of serum-free OptiMEM medium in a hematopoietic reconstituting ability in human and polystyrene tube and allowed to sit on the bench for at murine stem cells.26–29 least 20 min to form complexes. The liposome/DNA ␮ To date, most efforts have been made to improve the complex was then diluted to 250 l with serum-free Opti- efficiency of DNA packaging and its delivery into the cell MEM medium. Just before transfection, by development of new liposome formulations which supernatant was gently removed and cells were rinsed have also to be compatible with administration to human once with serum-free medium before being overlaid with ␮ subjects. It may now be necessary to direct some of our 250 l of the liposome/DNA complex. After 5-h incu- ° efforts toward understanding of the mechanisms which bation at 37 C, liposome/DNA complexes were aspirated render cells responsive to DNA transfer. and 1 ml of complete fresh medium (50% Dulbecco, 50% RPMI) was added. The cells were maintained at 37°C for 40 h until purification of the suspension cells and ␤- Materials and methods galactosidase assay. Serum and antibiotics were not present during lipid-mediated transfection. Cells KG1a,30 K562 and TF131,32 are human hematopoietic cells Purification of nonadherent transfected cells expressing the CD34 cell surface marker. Jurkat is a At 40 h after transfection, mixed cells were washed gently human CD4 T lymphocyte. Both human CD34 and CD4 once with phosphate-buffered saline (PBS), harvested by cell lines are grown in suspension. NIH3T3 and HeLa are trypsinization, resuspended in 2.5 ml of complete respectively a murine fibroblast and a human epithelial- medium and incubated for 2 h at 37°C in a 25-cm2 flask like cell line. MS5 is a murine stromal cell which has been to allow the adherent cells to attach to the bottom of the reported to support the growth and maintenance of the flask. The suspension cells were then gently collected, hematopoietic reconstituting ability of murine and loaded into a well of a six-well dish and incubated for 2 human blood progenitor cells.28,33,34 NIH3T3, MS5 and additional hours as before. This last procedure was Adhesion-assisted lipofection in hematopoietic cells H Keller et al 937 repeated once. The purity of the resulting cell suspension by two-color flow cytometry to detect expression of EGFP was estimated by FACS analysis. protein in CD4 and CD8 T lymphocytes.

X-gal staining procedure Harvesting cells and nuclei X-gal staining was performed according to the modified Cells (5 × 105) were harvested 48 h or 7 days after trans- standard procedure.35–37 Quantification of the transfected fection, washed once with PBS, resuspended in 500 ␮l cells was performed by counting in randomly selected washing buffer (140 mm NaCl, 1.5 mm MgCl2 10 mm Tris- fields the number of blue-stained cells under inverted HCI and 0.001% NP-40) and incubated for 5 min on ice. phase microscopy. The cells were washed once again with 500 ␮l washing buffer. Washing with NP-40 detergent was performed to RT-PCR analysis eliminate liposome particles and DNA plasmid molecules Total RNA was purified from 5 × 105 cells using the collected with culture medium supernatant or weakly trapped at the surface of the plasma membrane. Cells RNAPLUS (Quantum Bioprobe, Montreuil, France) pro- ␮ cedure. The aqueous phase was extracted twice with were then resuspended in 100 l lysis buffer (3 mm chloroform-isoamyl-alcohol before isopropanol precipi- CaCl2,2mm MgAc, 0.1 mm EDTA, 10 mm Tris-HCl pH tation. RNA samples were resuspended in 20 ␮l of water. 8.0, 1 mm DTT) supplemented with 0.2% NP-40 and incu- Two microliters of RNA were subjected to 1 unit of bated on ice for 10 min. A drop of the nuclei suspension DNase I in a 20 ␮l reaction as described in the GibcoBRL was examined microscopically to ensure that the nuclei ␮ were not ruptured and cells were lysed completely. The kit. A 2 l RNA sample was subjected to both DNase I ␮ and RNase A (1 ␮g/␮l). DNase I was inactivated by the nuclei solution was then gently layered on to 500 lofa ␮ 0.32 m sucrose cushion in lysis buffer and centrifuged at addition of 2 lof25mm EDTA solution to the reaction ° mixture and heating for 10 min at 65°C. Half of the pre- 4 C at 400 g for 5 min. While the cytoplasmic fraction was discarded, the nuclei pellet was gently resuspended in treated RNAs were retrotranscribed with SUPERSCRIPT ␮ II reverse transcriptase and 1 ␮g oligo-dt (GIBCO-BRL 500 l of lysis buffer and centrifuged once more. kit) in a 40 ␮l reaction supplemented with 3 mm MgCl . 2 ␤ Two microliters of single-stranded cDNAs were then Southern analysis of pCMV DNA content ␮ subjected to 20 cycles of amplification essentially as Whole cells or nuclei were gently resuspended in 285 l described (Perkin Elmer Applied Biosystems, Courta- extraction buffer (10 mm Tris-HCl pH 7.5, 0.1 m NaCl, ␮ boeuf, France) using Lac1 (5Ј-CGGACTCTAGAGGAT 10 mm EDTA). After addition of 15 l SDS 10%, the CCGGTACT) and Lac2 (5Ј-CAGTTTGAGGGGACG samples were shaken vigorously for 1 min at room tem- ACGACAGT) oligonucleotide primers. Southern analysis perature and extracted with phenol/chloroform. Total of RT-PCR products was performed using a 32P-labeled nucleic acids were recovered by ethanol precipitation. DNA probe specific for the ␤-galactosidase gene. Total nucleic acids were digested by EcoRV and NotI, electrophoresed in 1.0% agarose gel, transferred to nitro- cellulose and hybridized with a 32P-labeled ␤-galacto- and reagents sidase DNA probe. The hybridization signal correspond- 34-1-2S mouse antimouse MHC-H-2d monoclonal anti- ing to the resulting 1.2 kb DNA fragment was quantified body (mAb) (lgG2a)38 was kindly provided by Dr M Pla by phosphor-sensitive luminescence (Phosphorimager; (Hoˆpital Saint-Louis, France). Mouse anti-HLA class I Molecular Dynamics, Bondoufle, France) and compared (lgG2a, HLA-A,B,C; MAS 1532, Harlan Sera-Lab, Sussex, with serial dilutions of digested pCMV␤ DNA. UK), anti-CD34 (lgG1, QBEnd/10, Novo Castra, New- castle, UK), anti-CD4 (lgG1, kappa; MT310, DAKO, Trappes, France), anti-CD8 (lgG1, kappa; M707, DAKO), Acknowledgements lgG1 negative control (X-0931, DAKO) mAbs, as well as labeled goat antimouse mAbs (lgG H+L, Jackson Immuno This work was supported by the Centre National de la Research Laboratories, West Grove, PA, USA) conjugated Recherche Scientifique and the Institut Pasteur. We thank to fluorescein (FITC) or R-phycoerythrin (R-PE), were Jerry Bram for critically reading the manuscript. obtained from the manufacturers. References Flow cytometric analysis 1 Felgner PL et al. Lipofection: a highly efficient, lipid-mediated Cells (2.5 × 105) were pelleted in a Falcon tube (2052) and ␮ DNA transfection procedure. Proc Natl Acad Sci USA 1987; 84: resuspended in 100 l PBS supplemented with 0.5% BSA. 7413–7417. Cells were then incubated with specific mAbs for 20 min 2 Felgner PL, Ringold GM. Cationic liposome-mediated transfec- at 4°C followed by washing with 4 ml PBS-BSA. Washed tion. Nature 1989; 337: 387–388. cells were then incubated for 20 min at 4°C with FITC or 3 Felgner JH et al. Enhanced gene delivery and mechanism studies Ј R-PE-labeled goat anti-mouse F(ab )2 fragment. Finally, with a novel series of cationic lipid formulation. J Biol Chem the cells were washed once again with 4 ml PBS and 1994; 269: 2550–2561. resuspended in 0.5 ml PBS supplemented or not with 4 Stewart MJ et al. Gene transfer in vivo with DNA–liposome com- 0.5 ␮g/ml propidium iodide. Flow cytometric analysis of plexes: safety and acute toxicity in mice. Hum Gene Ther 1992; stained cells was performed with a FACScan flow cyto- 3: 267–275. 5 Hawley-Nelson P et al. Lipofectamine reagent: a new, higher meter and a CellQuest 1.2 software (Becton Dickinson, efficiency polycationic liposome transfection reagent. Focus Mountain View, CA, USA). Dead cells which stained 1993; 15: 73–79. with propidium iodide were excluded by prior electronic 6 Labat-Moleur F et al. An electron microscopy study into the gating. Analysis of mixed NIH3T3-depleted PBMC trans- mechanism of gene transfer with lipopolyamines. fected with liposome/pEGFP complexes was carried out 1996; 3: 1010–1017. Adhesion-assisted lipofection in hematopoietic cells H Keller et al 938 7 Harrison GS et al. Optimization of gene transfer using cationic 24 Dowty ME, Williams PW, Zhang G, Wolff JA. Plasmid DNA lipids in cell lines and primary human CD34+ and CD34+ hema- entry into post-mitotic nuclei of primary rat myotubes. Proc Natl topoietic cells. Bio Techniques 1995; 19: 816–823. Acad Sci USA 1995; 92: 4572–4576. 8 Aksentijevich I et al. In vitro and in vivo liposome-mediated gene 25 Bebok Z et al. Efficiency of plasmid delivery and expression after transfer leads to human MDR1 expression in mouse bone mar- lipid-mediated gene transfer to human cells in vitro. J Pharm Exp row progenitor cells. Hum Gene Ther 1996; 7: 1111–1122. Ther 1996; 279: 1462–1469. 9 Behr JP. Gene transfer with synthetic cationic amphiphiles: pros- 26 Kobari L et al. Hematopoietic promoting activity of the murine pects for gene therapy. Bioconj Chem 1994; 5: 382–389. stromal cell line MS-5 is not related to the expression of the 10 Zhou XH, Huang L. DNA transfection mediated by cationic major hematopoietic cytokines. J Cell Physiol 1995; 163: 295–304. liposomes containing lipopolylysine characterization and mech- 27 Xu LC et al. Long-term in vivo expression of the human glucocer- anism of action. Biochim Biophys Acta 1994; 1189: 195–203. ebrosidase gene in nonhuman primate after CD34+ hematopo- 11 Friend DS, Papahadjopoulos D, Debs RJ. Endocytosis and intra- ietic cell transduction with cell-free retroviral vector prep- cellular processing accompanying transfection mediated by cat- arations. Proc Natl Acad Sci USA 1995; 92: 4372–4376. ionic liposomes. Biochim Biophys Acta 1996; 1278: 41–50. 28 Goncalves F et al. Stromal cells maintain the radioprotective 12 Behr JP, Demeneix B, Loeffler JP, Perez-Mutul J. Efficient gene capacity of CFU-S during retroviral infection. Gene Therapy 1996; transfer into mammalian primary endocrine cells with lipopoly- 3: 761–768. amine-coated DNA. Proc Natl Acad Sci USA 1989; 86: 6982–6986. 29 Gordon MY. Stromal cells negatively regulate primitive haemo- 13 Barthel F, Remy JS, Loeffler JP, Behr JP. Gene transfer optimiz- poietic progenitor cell activation via phosphatidyl inisitol-anch- ation with lipospermine-coated DNA. DNA Cell Biol 1993; 6: ored cell adhesion/signaling mechanism. Br J Haematol 1997; 96: 553–560. 647–653. 14 Zabner J et al. Cellular and molecular barrier to gene transfer 30 Koeffler HP et al. An undifferentiated variant derived from the by a cationic lipid. J Biol Chem 1995; 270: 18997–19007. human acute myelogenous leukemia cell line (KG-1). Blood 1980; 15 Farhood H, Serbina N, Huang L. The role of phosphatidylethan- 56: 265–273. olamine in cationic liposome mediated gene transfer. Biochim 31 Kitamura T et al. Establishment and characterization of a unique Biophys Acta 1995; 1235: 289–295. human cell line that proliferates dependently on GM-CSF, IL-3 16 Fasbender A et al. Effect of co-lipids in enhancing cationic lipid- or erythropoietin. J Cell Physiol 1989; 140: 323–334. mediated gene transfer in vitro and in vivo. Gene Therapy 1997; 32 Bagnis C et al. Retroviral transfer of the nlsLacZ gene into 4: 716–725. human CD34+ cell populations and into TF-1 cells: future pros- 17 Moritz T, Patel VP, Williams DA. Bone marrow extracellular matrix molecules improve gene transfer into human hematopo- pects in gene therapy. Hum Gene Ther 1994; 5: 1325–1333. ietic cells via retroviral vectors. J Clin Invest 1994; 93: 1451–1457. 33 Itoh K et al. Reproducible establishment of haemopoietic sup- 18 Xu LC et al. Growth factors and stromal support generate very portive stromal cell lines from murine bone marrow. Exp Hema- efficient retroviral transduction of peripheral blood CD34+ cells tol 1989; 17: 143–153. 34 Issaad C et al. A murine stromal cell line allows the proliferation from Gaucher patients. Blood 1995; 86: 141–146. + − 19 Moore KA et al. Stromal support enhances cell-free retroviral of very primitive human CD34 , CD38 progenitor cells in long- vector transduction of human bone marrow long-term culture- term cultures and semisolid assays. Blood 1993; 81: 2916–2924. initiating cells. Blood 1992; 79: 1393–1399. 35 MacGregor GR, Mogg AE, Burke JF, Caskey CT. Histochemical 20 Bordignon C et al. Gene therapy in peripheral blood lympho- staining of clonal mammalian cell lines expressing E. coli beta cytes and bone marrow for ADA-immunodeficient patients. galactosidase indicates heterogeneous expression of the bac- Science 1995; 270: 470–475. terial gene. Somat Cell Mol Genet 1987; 13: 253–265. 21 Hanenberg H et al. colocalization of retrovirus and target cells 36 Lin WC, Culp LA. Selectable plasmid vectors with alternative on specific fibronectin fragments increases genetic transduction and ultrasensitive histochemical marker genes. Bio Techniques of mammalian cells. Nature Med 1996; 2: 876–882. 1991; 11: 344–351. 22 Moritz T et al. Fibronectin improves transduction of reconstitut- 37 Bagnis C et al. Leukemogenicity of v-myb transformed monob- ing hematopoietic stem cells by retroviral vectors: evidence of lasts cells can be modulated by normal bone marrow environ- direct viral binding to chymotryptic carboxy-terminal frag- ment. Oncogene 1993; 8: 737–743. ments. Blood 1996; 88: 855–862. 38 Ozato K, Mayer NM, Sachs DH. Monoclonal antibodies to 23 Logan JL et al. Cationic lipids for reporter gene and CFTR trans- mouse major histocompatibility complex antigens. Transplan- fer to rat pulmonary epithelium. Gene Therapy 1995; 2: 38–49. tation 1982; 34: 113–120.