|Home | About | Journals | Submit | Contact Us | Français|
Mesenchymal stem cells (MSCs) have attracted much attention as potential platforms for transgene delivery and cell-based therapy for human disease. MSCs have the capability to self-renew and retain multipotency after extensive expansion in vitro, making them attractive targets for ex vivo modification and autologous transplantation. Viral vectors, including lentiviral vectors, provide an efficient means for transgene delivery into human MSCs. In contrast, mouse MSCs have proven more difficult to transduce with lentiviral vectors than their human counterparts, and because many studies use mouse models of human disease, an improved method of transduction would facilitate studies using ex vivo-modified mouse MSCs. We have worked toward improving the production of human immunodeficiency virus type 1 (HIV-1)-based lentiviral vectors and optimizing transduction conditions for mouse MSCs using lentivirus vectors pseudotyped with the vesicular stomatitis virus G glycoprotein (VSV-G), the ecotropic murine leukemia virus envelope glycoprotein (MLV-E), and the glycoproteins derived from the Armstrong and WE strains of lymphocytic choriomeningitis virus (LCMV-Arm, LCMV-WE). Mouse MSCs were readily transduced following overnight incubation using a multiplicity of infection of at least 40. Alternatively, mouse MSCs in suspension were readily transduced after a 1-h exposure to lentiviral pseudotypes immediately following trypsin treatment or retrieval from storage in liquid nitrogen. LCMV-WE pseudotypes resulted in efficient transduction of mouse MSCs with less toxicity than VSV-G pseudotypes. In conclusion, our improved production and transduction conditions for lentiviral vectors resulted in efficient transduction of mouse MSCs, and these improvements should facilitate the application of such cells in the context of mouse models of human disease.
For transgene delivery into mesenchymal stem cells (MSCs), viral vector systems including adenoviral vectors, vectors based on adeno-associated virus (AAV), and retroviral vectors are commonly used . The ability of MSCs to self-renew at a high proliferation rate led to the prediction that they would be ideal targets for transgene delivery strategies involving retroviral vectors. However, a major limitation of transduction approaches involving oncogenic retroviral vectors. such as Moloney murine leukemia virus, is a general lack of long-term transgene expression [2,3]. Vectors based on murine stem cell virus appear to be less prone to transcriptional silencing of gene expression, and thus appear to be more promising .
Recent results from several labs have indicated that human immunodeficiency virus type 1 (HIV-1)-based vectors are very efficient at delivering and expressing transgenes in human MSCs [5–8]. For example, a single round of transduction using unconcentrated HIV-1-based lentiviral vectors led to the efficient transduction of human MSCs and sustained transgene expression up to at least 5 months . An advantage of lentiviral vectors over vectors based on oncogenic retroviruses is their ability to transduce nondividing cells . This property is important given that a relatively large subset (20%) of mesenchymal progenitor cells (MPCs) has been described to be quiescent .
Transgene delivery strategies using lentiviral vectors involving MSCs from other species have also been reported. For example, Lee et al.  have used self-inactivating HIV-1-based lentiviral vectors to transduce MSCs derived from fetal rhesus monkey bone marrow. Flow cytometric analyses indicated an 8- to 10-fold greater quantity of green fluorescent protein (GFP)-expressing rhesus MSCs when cells were transduced with vectors bearing the cytomegalovirus immediate early (CMV) or translation elongation factor-1α (EF-1α) promoters as compared to vectors bearing the phosphoglycerate kinase (PGK) promoter. Transduced rhesus MSCs differentiated toward an osteogenic lineage comparable to untransduced MSCs. In agreement with the reports published by Zhang et al. , these findings suggested that HIV-1-derived lentiviral vectors can efficiently transduce rhesus MSCs in vitro without inhibiting their differentiation potential.
In a recent report, McMahon et al.  performed a direct comparison of different vectors on rat MSCs, including lentiviral vectors, vectors based on adenovirus or AAV, as well as nonviral vectors. The results indicated that VSV-G pseudotyped HIV-1-based vectors were the vector of choice for rat MSCs. Furthermore, it was shown that the transduction process and the high levels of reporter gene expression achieved did not have any deleterious effect on the ability of the cells to differentiate down the adipogenic pathway. Efficient lentivirus-mediated gene transfer into mouse MSCs has been more challenging, in part because of host range barriers for HIV-1 in such cells, including tissue-specific restriction factors  and blocks in the nuclear uptake of the preintegration complex .
In this study, we present improved protocols for the transduction of mouse MSCs with pseudotyped lentiviral vectors that result in significant transduction efficiency and reduced cell toxicity. We investigated alternative transduction conditions, such as varying transduction length and transducing cells immediately following trypsin treatment or retrieval from storage in liquid nitrogen. We also considered aspects of lentivirus production and concentration to maximize viral titers and minimize potential contaminants of vector stocks. Investigation of alternative pseudotypes led us to develop and test a modified titration method based on quantifying lentiviral RNA content of vector preparations. Using this titration method to normalize vector quantity, we compared the ability of vectors pseudotyped with vesicular stomatitis virus G glycoprotein (VSV-G), murine leukemia virus envelope glycoprotein (MLV-E), and glycoproteins derived from the Armstrong and WE strains of lymphocytic choriomeningitis virus (LCMV-Arm and LCMV-WE).
Murine MSCs were harvested from the tibias and femurs of C57/BL6 mice, as previously described  and provided by the Tulane Center for Gene Therapy. The plastic-adherent population was grown in complete culture medium, consisting of Iscove's modified Dulbecco's medium (IMDM; Invitrogen, Carlsbad, CA), 10% fetal bovine serum (FBS; Atlanta Biologicals, Lawrenceville, GA), 10% horse serum (HS; Hyclone, Logan, UT), 2 mM l-glutamine (Invitrogen), and 100 U/ml penicillin/streptomycin (Pen/Strep; Invitrogen) and passaged at low density (50 cells/cm2). For long-term storage, cells were resuspended in IMDM with 20% FBS, 20% HS, 5% dimethylsulfoxide (DMSO; Sigma-Aldrich, St. Louis, MO), frozen at −1°C/min until reaching −80°C, and transferred to liquid nitrogen. Passage-five cells were used for transduction experiments.
To construct the pNL-EGFP/CMV/WPREΔU3 lentiviral vector plasmid, a 591-bp woodchuck hepatitis virus post-transcriptional element (WPRE) was added between the unique Xho I and Kpn I sites, downstream of the enhanced green fluorescent protein (EGFP) coding region present in pNL-EGFP/CMV/ΔU3  (see http://www.medschool.lsuhsc.edu/reiser). The WPRE fragment was generated by PCR amplification using pWHV8 (ATCC, Manassas, VA) as a template. Primers used were WPRE-S (5′-AAC TCG AGA ATC AAC CTC TGG ATT ACA A-3′) and WPRE-A (5′-AAG GTA CCC AGG CGG GGA GGC GGC CCA A-3′). The pNL-EGFP/EF-1α/WPREΔU3 plasmid was constructed by PCR amplifying the EF-1α/human T cell leukemia virus (HTLV) type 1 long terminal repeat (LTR) hybrid promoter from pGT70LacZ (Invivogen, San Diego, CA) and inserting it between the Hinc II and Nhe I sites of pNL-EGFP/CMV/WPREΔU3. The pNLhCXCR2/EF-1α/WPREΔU3 plasmid was constructed by excising the human CXCR2 open reading frame (ORF) from pCRII-hCXCR2 (kindly provided by Dr. Tim Sparer)  and subcloning it into pNL-EGFP/EF-1α/WPREΔU3.
Lentiviral vectors pseudotyped with the VSV-G  MLV-E , LCMV-Arm , and LCMV-WE glycoproteins  were generated by calcium phosphate-mediated transfection of 293T cells with modifications as described [20,21]. 293T cells were plated in 150-cm2 plates at a density of 8 × 106 cells in 25 ml of Dulbecco's modified Eagle's medium (DMEM, high-glucose) supplemented with 10% FBS, 1% Glutamax (Invitrogen), 1% Pen/Strep, with or without 0.3% HyQ LipiMate (Hyclone), Chemically Defined Lipid Concentrate (Invitrogen), or cholesterol (Sigma). Twenty four hours later, chloroquine (Sigma) was added to the medium at a final concentration of 25 μM. Lentivirus vector, packaging, and envelope glycoprotein plasmids were mixed together in 3 ml of 0.25 M CaCl2 per plate, added to 3 ml of 2× HEPES-buffered saline (HBS) under gentle vortexing, and pipetted into the medium. The amount of DNA used per plate was 21 μg of lentiviral vector plasmids pNL-EGFP/EF-1α/WPRE ΔU3, pNL-EGFP/CMV/WPREΔU3, or pNL-hCXCR2/EF-1α/WPRE Δ, 14 μg of packaging plasmid pCD/NL-BH*ΔΔΔ , and either 7 μg pLTR-G  or 21 μg pCAGGS-LCMV-WE, pCAGGS-LCMV-Arm (provided by Dr. Juan Carlos de la Torre, The Scripps Research Institute), or pLTR-MLV-E. The pLTR-MLV-E plasmid was derived from the pLTR 4070A Env plasmid  by replacing the 4070A Env sequence with a fragment encoding the MLV ecotropic Env. The medium was removed 16 h after transfection and replaced with 17 ml of fresh DMEM, 10% FBS, and 1% Glutamax. Forty eight hours after transfection, the vector-containing medium was collected and spun at 500 × g for 5 min, filtered through a 0.45-μ pore size filter (Corning, Corning, NY) and stored at –80°C. For vector concentration by ultracentrifugation, the vector-containing medium from two plates was underlaid with 4 ml of 20% sucrose and centrifuged for 2 h at 25,000 rpm, 4°C using a Beckman SW28 ultracentrifuge rotor (Beckman Coulter, Fullerton, CA). The resulting pellet was dissolved in 100 μl of PBS without calcium or magnesium (Invitrogen) for 2 h at 4°C. Vector aliquots were stored at –80°C.
Four milliliters of vector-containing medium or 100 μl of concentrated vectors diluted in 3.9 ml of PBS were loaded into Beckman SW28 UltraClear tubes (Beckman Coulter) and underlaid with a continuous gradient of 10–30% OptiPrep (Axis-Shield, Oslo, Norway) in 20 mM Tris-HCl, 1 mM EDTA, and 0.85 (wt/vol) NaCl, pH 7.5. The gradient was spun at 25,000 rpm, for 4 h at 4°C, using an SW28 rotor. Eighteen 2-ml fractions were collected, and each fraction was titrated by end-point dilution on human osteosarcoma (HOS) cells , or measured for p24 antigen levels using a HIV-1 p24 Antigen ELISA kit (Zeptometrix, Buffalo, NY) .
For lentiviral vector concentration using Mustang Q Acrodisks (PALL, East Hills, NY), vector-containing supernatants were adjusted to 25 mM Tris-HCl, pH 8.0, 0.3 M NaCl and loaded onto a Mustang Q Acrodisk (bed volume 0.18 ml). The membrane was washed with loading buffer, and the flowthrough was discarded. Vectors were eluted with 10 ml of 25 mM Tris-HCl, pH 8.0, and 1.5 M NaCl directly into 25 ml of PBS. To concentrate the vectors further and to remove any residual salt, the diluted eluate was loaded into a Beckman SW28 UltraClear tube and centrifuged at 25,000 rpm for 2 h at 4°C and the pellets were resuspended in PBS as described above.
Quantitative real-time PCR analysis of proviral copy numbers was similar to that reported before . The sequences of the WPRE-specific primers were: 5′-CCT TTC CGG GAC TTT CGC TTT-3′ (forward primer); 5′-GCA GAA TCC AGG TGG CAA CA-3′ (reverse primer), and 5′-FAM-ACT CAT CGC CGC CTG CCT TGC C-TAMRA-3′ (probe). The cycling conditions were 10 min at 95°C, then 40 cycles of 95°C for 15 sec and 60°C for 1 min. Genomic vector copies in each sample were normalized to human RNaseP gene copies using specific primers and probes (TaqMan DNA Template Reagent Kit, Applied Biosystems, Foster City, CA)
A 2μl lentiviral vector sample was diluted into 398 μl of 25 mM Tris-HCl, pH 8.0, and treated with 50 pg of RNase A (USB, Cleveland, OH) at 37°C for 10 min. Then 20 μl of SUPERase-In (Ambion, Austin, TX) was added prior to performing RNA extraction using the PureLink Viral RNA/DNA kit (Invitrogen). The final elution volume was 30 μl. Reverse transcription PCR was performed using an iScript cDNA synthesis kit (BioRad, Hercules, CA) on half of the RNA. The other half was used in control reactions lacking reverse transcriptase. Five microliters of each reaction was used for quantitative real-time PCR as described.
For transduction comparisons using preplated, trypsinized, and thawed mouse MSCs, preplated cells were plated 24 h before transduction. Trypsinized cells were detached using 0.25% trypsin and 1 mM EDTA (Invitrogen), inactivated with complete culture medium, washed, and counted immediately before transduction. Thawed cells were removed from liquid nitrogen, thawed rapidly, washed with complete culture medium, counted, and used directly for transduction. Equal numbers of cells were used for each experiment. Transductions were performed in 0.5 ml of transduction medium (IMDM +10% heat-inactivated FBS) in the presence of 8 μg/ml Polybrene (Sigma), and medium was changed to complete culture medium 1 or 16 h after addition of lentivirus.
Transduced cells expressing EGFP were trypsinized, inactivated with complete culture medium, washed twice with PBS containing 2% FBS, and analyzed using a FACSCalibur system (BD Biosciences, San Jose, CA). For antibody staining of hCXCR2-transduced cells, cells were resuspended in PBS and 2% FBS containing either phycoerythrin (PE)-labeled anti-human CXCR2 monoclonal antibody (Clone 48311, R&D Systems, Inc., Minneapolis, MN) or mouse immunoglobulin G2A (IgG2A) PE-labeled isotype control (Clone 20102, R&D) and incubated for 30 min at 4°C. Cells were washed twice with PBS and 2% FBS and analyzed by flow cytometry.
To assay the proliferation rate of MSCs that were transduced overnight, 100 viable cells were added to six-well plates (Corning) 48 h before transduction. Following overnight transduction, the medium was removed, the cells were washed once with PBS, and the wells were filled with 2 ml of freshly prepared complete culture medium. The plates were then placed in a 37°C, 5% CO2 humidified incubator for 14 days, after which the medium was removed and the wells washed with PBS. Colonies were stained for 10 min with 3.0% crystal violet in 100% methanol, and after washing the wells three times with deionized water, the colonies that were 1 mm or larger in diameter were counted. Assay of MSCs transduced for 1 h was performed as described, with the exception that cells were transduced in suspension after trypsin treatment and subsequently plated into 10-cm dishes (Corning) filled with 10 ml of medium.
Statistical analyses were performed using one-way analysis of variance (ANOVA) followed by multiple paired comparisons (Student's t-test). * represents p < 0.05; ** represents p < 0.01; *** represents p < 0.001.
To increase the participation of MSCs in site-directed tissue repair, we are using lentiviral vector-mediated gene delivery strategies to modify MSC trafficking and achieve higher rates of engraftment in damaged tissues.
Previously, we have described conditions allowing efficient lentiviral vector-mediated transgene delivery into human MSCs, and we revealed that efficient lentiviral vector-mediated gene transfer was possible using a multiplicity of infection (MOI) of 1 [5,7]. At similar low MOIs, transduction of mouse MSCs was found to be inefficient, consistent with the view that there are impediments to efficient lentiviral vector-mediated transduction of murine cells. However, at MOIs of 40 and 80, up to 50.3% and 60.5% of the cells were GFP+ following exposure to the lentiviral vectors overnight (Fig. 1A). We also observed modest transduction of mouse MSCs following exposure to the vectors for just 1 h (Fig. 1B). Furthermore, transduction strategies using suspended mouse MSCs immediately after retrieval of such cells from storage in liquid nitrogen or after trypsin treatment were significantly more efficient as those using preplated MSCs (Fig. 2). Thus, there does not appear to be a need for MSCs to be plated prior to transduction, and using MSCs in suspension may increase transduction efficiency.
Because efficient transduction of mouse MSCs requires high MOIs, it is important to improve lentivirus yields. Several factors that affect the stability and/or yield of retroviral vectors have previously been reported. For example, pharmacological disruption of lipid rafts and cholesterol depletion of producer cells has been shown to interfere with virus particle formation . We surmised that decreased cholesterol and lipid levels of producer cells is a limiting step in lentiviral vector production and thus evaluated the effects of various lipid additives on the relative titers of such vector stocks. The results presented in Fig. 3A show significant increases in vector titers as a result of lipid additives during vector production. Vector titers on HOS cells increased 3.7-fold in the presence of cholesterol, 4.8-fold in the presence of chemically defined lipid concentrate, and 11.2-fold in the presence of LipiMate, relative to vector stocks produced in the absence of lipid additives. Similar increases in vector titers as a result of lipid additives were seen with mouse MSCs (Fig. 3B).
We next tested the ability of lentiviral vectors bearing glycoproteins other than VSV-G to transduce mouse MSCs. Park et al.  have demonstrated that lentiviral vectors pseudotyped with the LCMV-WE glycoprotein displayed reduced liver toxicity in mice compared to VSV-G pseudotypes. These findings prompted us to investigate alternative envelopes, including LCMV-WE and LCMV-Arm pseudotypes [18,19]. Preliminary titration of alternative pseudotypes based on p24 levels generated ambiguous results, possibly caused by the existence of unincorporated p24 in the vector stocks. To determine the relationships between p24 levels and transducing units in more detail, we subjected a crude lentivirus vector preparation to rate-zonal ultracentrifugation using a 10– 30% gradient of OptiPrep (Fig. 4A). A total of 18 fractions were collected and individual fractions were tested for vector transducing units and p24 content using a p24 enzyme-linked immunoassay (ELISA) test. The results presented in Fig. 4A show that a relatively large fraction of the total p24 did not co-sediment with infectious vector particles. However, concentrating the lentivirus preparation prior to rate-zonal ultracentrifugation improved the correlation between p24 levels and transducing units (Fig. 4B). These results indicate that titer determinations based on p24 content are unreliable for crude vector stocks but may be applicable for vectors concentrated by ultracentrifugation over a sucrose cushion. As an alternative to p24-based titration, we used a modified version of the protocol based on virion RNA originally described by Sastry et al.  to adjust titers of lentiviral vectors bearing alternative envelope glycoproteins.
The results presented in Fig. 5 show the transduction efficiencies of lentiviral vectors pseudotyped with various envelope glycoproteins including VSV-G, MLV-E, LCMV-Arm, and LCMV-WE. In all cases, titers were adjusted by quantitative RT-PCR based on virion RNA. A total of 7.9 × 109 virus particles were used to transduce 5 × 104 mouse MSCs. It is evident that VSV-G pseudotypes were most efficient both during short-term transduction (1 h) and after overnight transduction (16 h). However, vectors pseudotyped with the LCMV-WE glycoprotein were also efficient, particularly after overnight transduction (Fig. 5).
Table 1 presents our findings regarding the toxicity of VSV-G-pseudotyped vectors before and after Mustang Q anion-exchange chromatography and compared to MLV-E, LCMV-Arm, and LCMV-WE pseudotypes. Both Mustang Q chromatography and pseudotyping with alternative envelopes resulted in higher colony formation than VSV-G, indicating less toxicity to progenitor cells. We also found that transduction for 1 h did not result in any differences in colony formation between groups, indicating a low toxicity associated with short exposure.
Having optimized lentiviral vector production and transduction for mouse MSCs, we validated our findings with a transgene encoding a cell-surface protein. Expression of human chemokine receptor CXCR2 was assessed in mouse MSCs following lentiviral vector-mediated gene transfer. The results shown in Fig. 6 demonstrate that, using our improved transduction conditions, up to 88.4% of mouse MSCs expressed hCXCR2 after being transduced overnight at an MOI of 90.
In an attempt to optimize HIV-1-based lentiviral vectors for efficient transgene delivery into mouse MSCs, we have investigated various transduction protocols and have tested several vector pseudotypes. In general, mouse cells do not support HIV-1 replication because of host range barriers at various steps including virus entry, nuclear import , RNA splicing , polyprotein processing, assembly, and release. For example, Noser et al.  reported that HIV-1 infection of murine cells is inhibited by dominant factors related to immunophilins and that competitive inhibitors of cyclophilins, including cyclosporin and the related compound Debio-025, stimulated HIV-1 vector transduction of primary murine bone marrow-derived cells and macrophages up to 20-fold. On the basis of these findings we tested the impact of cyclosporin A on mouse MSCs and did not observe improvements in transduction efficiencies (Ricks, unpublished).
Our results indicate that lentivirus-mediated transduction of mouse MSCs was robust, provided that high enough MOIs were used. We were able to boost vector titers substantially by adding lipids such as LipiMate during virus production. Increased vector titers were observed both using HOS cells (Fig. 3A) and mouse MSCs (Fig. 3B). This is consistent with the view that lipid additives primarily affected virus production. It was is also evident from the results shown in Fig. 2 that transduction efficiencies involving cells in suspension were significantly higher than those involving adherent cells. However, it was apparent from the CFU assay displayed in Table 1 that overnight transductions at high MOIs resulted in VSV-G-mediated cell toxicity . This toxicity issue could be partially overcome by reducing the time of vector exposure to 1 h, or by using alternative pseudotypes including vectors bearing the LCMV-WE glycoprotein. These findings are in line with those reported earlier by Park et al. , which showed that LCMV-pseudotyped lentiviral vectors resulted in reduced systemic or hepatic injury compared to VSV-G pseudotypes after in vivo administration.
Our attempts to compare the transduction efficiencies of the various pseudotypes tested revealed two technical issues in quantitating viral titers: (1) assay selection, and (2) cell line transduction efficiency. Different methods have been used to determine lentiviral vector titers, including measures based on the number of vector particles present in a virus stock and measures derived from the number of proviral copies in transduced target cells. Virus particle numbers can be determined using real-time PCR based on strong-stop cDNA present in virions . Alternatively, the amount of a virus core protein present in the vector preparation, such as p24 Gag, is determined by ELISA to arrive at relative particle titers . Functional titration assays are based on vector-encoded reporter gene expression. For example, vectors encoding GFP have been titrated using fluorescence-activated cell sorting (FACS) analysis [5,30]. For vectors that do not contain a reporter gene, proviral DNA copy numbers determined by real-time PCR using DNA extracted from transduced cells have been used to measure titers . The cell line used for titration is significant, as receptors for a given pseudotype may vary among cell lines, possibly producing a falsely depressed titer .
To adjust titers of pseudotyped vectors, we decided to pursue particle-based titration methods. The results presented in Fig. 4A show that the correlation between p24 levels and transduction units was poor for unconcentrated vector stocks. Thus, this avenue was abandoned in favor of a modified particle assay that is based on virion RNA . The performance of titer-adjusted vectors including VSV-G, MLV-E, LCMV-Arm, and LCMV-WE pseudotypes is displayed in Fig. 5. It is evident from this analysis that VSV-G pseudotypes were the most efficient followed by LCMV-WE pseudotypes. An attractive feature of LCMV-WE pseudotypes is that they appear to be less toxic as judged by the CFU assay.
We have also investigated a scaleable protocol for lentiviral vector concentration based on strong anion exchange membranes, such as Mustang Q Acrodisks, that allows viral concentration with little volume limitation and improved purification from serum proteins and cellular contaminants (Kutner, unpublished). Mustang Q-treated VSV-G pseudotypes were less toxic compared to VSV-G pseudotypes that had been concentrated by ultracentrifugation, as judged by the CFU assay.
In conclusion, our results show that HIV-1-based vectors appear to be efficient for delivering and expressing transgenes in mouse MSCs, provided that high MOIs and transduction protocols involving suspended cells are used. Our results also indicate that lipid additives such as LipiMate helped boost vector titers, thus simplifying the production of high-titer vector stocks. Finally, cell toxicity was lowest with LCMV-WE pseudotypes. Thus, for optimal transduction of mouse MSCs we recommend the use of LCMV-WE pseudotypes produced in the presence of LipiMate.
The preclinical utility of genetically modified mouse MSCs and their progenitors in the context of mouse models of human disease requires stable and long-term expression of the desired gene product as well as regulation of gene expression according to disease status. These goals may be achieved using lentiviral vectors.
We are grateful to Dr. Juan Carlos de la Torre for providing us with plasmids encoding LCMV-WE and LCMV-Arm glycoproteins and to Dr. Tim Sparer for providing the plasmid encoding hCXCR2. We thank Connie Porretta for assistance with FACS. Some of the materials employed in this work were provided by the Tulane Center for Gene Therapy through a grant from the National Center for Research Resources (NCRR) of the National Institutes of Health (NIH), grant P40RR017447. This work was supported by NIH grants NS044832, HL075161, and HL073770.