Human Fetal Mesenchymal Stem Cells as Vehicles for Gene Delivery



First-trimester fetal blood contains a readily expandable population of stem cells, human fetal mesenchymal stem cells (hfMSCs), which might be exploited for autologous intrauterine gene therapy. We investigated the self-renewal and differentiation of hfMSCs after transduction with onco-retroviral and lentiviral vectors. After transduction with either a MoMuLV retrovirus or an HIV-1-based lentiviral vector carrying the β-galactosidase and green fluorescent reporter gene, respectively, transgene expression, self-renewal, and differentiation capabilities were assessed 2 and 14 weeks later. Transduction with the lentiviral vector resulted in higher efficiencies than with the MoMuLV-based vector (mean, 97.7 ± 1.4% versus 80.2 ± 5.4%; p = .02). Transgene expression was maintained with lentiviral-transduced cells (94.6 ± 2.6%) but decreased over 14 weeks in culture with onco-retroviral-transduced cells (48.3 ± 3.9%). The self-renewal capability of these cells and their ability to undergo osteogenic, adipogenic, and myogenic differentiation was unimpaired after transduction with either vector. Finally, clonal expansion of lentivirally modified cells was expanded over 20 population doublings with maintenance of multiline age differentiation capacity. These results suggest that hfMSCs may be suitable targets for ex vivo genetic manipulation with onco-retroviral or lentiviral vectors without affecting their stem cell properties.


Genetic diseases with intrauterine onset can lead to permanent neural or musculoskeletal damage in early life. Examples include skeletal dysplasias, mucopolysaccharidoses, and other inborn errors of metabolism and some muscular dystrophies. To prevent this, stem cells have been proposed as targets for fetal gene therapy because of their ability to expand and give rise to many multipotent cell types. Intra-uterine treatment capitalizes on the immune naiveté of the developing fetus and offers the possibility of treating the disease before pathology manifestation.

Intrauterine transplantation has been attempted in several human pregnancies using allogeneic hemopoietic stem cells (HSCs), including maternal bone marrow (BM) for Rh alloimmunization and Chediak-Higashi disease, paternal BM for leukodystrophy, and fetal liver for bare lymphocyte syndrome. However, most cases resulted in no engraftment of transplanted cells, with the few well-documented successes being limited to fetuses with immunodeficiency syndromes [1]. Suggested reasons for these failures include the relatively late gestational age at which they were attempted (after 14 weeks), the small cell dose used [2], and, more likely, immune rejection of allogeneic cells.

The use of an autologous source of stem cells should largely overcome the immunological barrier. However, HSCs may not be good candidates, because studies in animal models using predominantly retroviral vectors have found limited transduction and subtherapeutic expression in hemopoietic stem cells because of both their quiescence and their low levels of the amphotropic receptor required for onco-retroviral entry [3]. An alternative target is mesenchymal stem cells (MSCs), which divide rapidly and, because of their high amphotropic receptor levels, are readily transducible with integrating vectors and maintain transgene expression in vitro and in vivo without affecting multipotentiality [4, 5]. As a result, MSC-based cell replacement and gene therapy are presently under investigation for therapeutic applications in myocardial [6]and neural [7] injury, tissue engineering [8], and genetic deficiency states such as enzyme deficiency syndromes [9] and muscular dystrophy [10]. Use of MSCs should obviate the rejection associated with intrauterine HSC transplantation, because MSCs have been shown to downregulate certain immune responses [11] and allogeneic/xenogeneic MSCs engraft in a wide range of organs in immunocompetent animal models [1214].

We recently reported that human fetal MSCs (hfMSCs) can be readily isolated and expanded from first-trimester blood, liver, and BM. Under permissive conditions, they differentiate into fat, bone, and cartilage [15], and recent work suggests they can also differentiate into neural [16] and muscle [17] cells. In pilot experiments, hfMSCs were transducible with a retroviral vector without affecting short-term self-renewal [18]. hfMSCs do not express class II major histocompatability antigens and do not elicit alloreactive lymphocyte proliferation in vitro [19, 20]. Transplantation in utero into a xenogeneic fetal sheep model resulted in widespread chimerism [21].

In this study, we transduced hfMSCs efficiently with both MoMuLV-derived and HIV-1-derived vectors. We achieved long-term transgene expression with either vector type, albeit to a lesser extent with MoMuLV, without affecting either self-renewal or multiline age differentiation capacity up to 14 weeks after transduction. These results highlight the potential utility of hfMSCs for autologous ex vivo gene therapy.

Materials and Methods


Blood and fetal tissue collection was approved by the Research Ethics Committee (Hammersmith and Queen Charlotte's Hospitals) in compliance with national guidelines regarding the use of fetal tissue for research purposes (Polkinghorne). All women gave written informed consent for collection and use of human tissues.

Sample Collection

First-trimester fetal blood samples (40–400μl)wereobtained by ultrasound-guided cardiac aspiration (n = 3) (crown-rump length of 24, 30, and 40 mm, corresponding to 9+1, 9+6, and 10+6 weeks of gestation) before clinically indicated surgical termination of pregnancy. Fetal gestational age was determined by crown-rump length measurement on ultrasound.

Culture of MSCs from Fetal Blood

Fetal blood was plated in 100mm dishes at 105 nucleated cells per ml [15] and cultured in D10, defined as 10% fetal bovine serum (FBS) (Stem Cell Technologies, Vancouver, Canada) in Dulbecco's modified Eagle's medium (DMEM) (Sigma-Aldrich Company Ltd, Dorset, U.K.) supplemented with 2 mmol/l L-glutamine, 50 IU/ml penicillin, and 50 mg/ ml streptomycin (Gibco BRL Life Technologies Ltd, Pais-ley, UK) at 37°C in 5% CO2. After 3 days, nonadherent cells were removed and the medium was replaced. Adherent cell colonies were detached with 0.25% trypsin EDTA (Stem Cell Technologies), expanded, cultured to confluence in 75-cm2 flasks, trypsinized, and stored in liquid nitrogen. hfMSCs at passages 3–7 were used in transduction experiments. We have previously reported that hfMSCs form an average of 8.2 colony-forming units (CFU) per 106 mononuclear cells plated (70 CFU/100 ml of fetal blood) [15].

Retroviral Production

For the generation of onco-retroviral vector, the plasmid pBMN-Z-I-Neo, which allows expression of LacZ and coexpression of a neomycin resistance gene via the encephalomyocarditis virus internal ribosome entry site (IRES) (Gift from G. Nolan, Stanford), was acquired together with a second-generation retrovirus producer line (SD 3443, ATCC, Manassas, VA) to generate helper-free amphotropic retroviruses. 293T producer cell lines were transiently transfected with 30 μg of plasmid via calcium phosphate precipitation (Profection, Promega, Southampton, UK). Supernatant was collected at 48 hours after transfection, filtered through a 0.2-μm filter for use in transduction experiments [22]. Retroviral titer approximated 1 × 106 viral particles per milliliter.

Low-passage hfMSCs were plated at 4 × 104/cm2 on 24-well plates overnight and then incubated with 1:1 retroviral supernatant: D10 in the presence of 4 μg/ml polybrene for 6 hours (multiplicity of infection [MOI], 12.5). Thereafter, cells were incubated with D10. A second or third cycle was performed on some samples 24 hours apart. In some experiments, samples were transduced under centrifugation at 1,000g at 25°C for 1 hour, followed by 5 hours of static transduction [5]. Control hfMSCs were set up in parallel with the addition of supernatant from mock-transfected 293T producer cells.

Analysis of transduction efficiency was performed 48 hours after transduction and at passage 8 and 24 via X-Gal staining as described below. Antibiotic selection with 1 mg/ ml G418 (Geneticin, Invitrogen Ltd., U.K.) selection over 6 days was used to achieve 95% of transduced cells expressing the transgene product β-galactosidase.

Lentiviral Production

The vesicular stomatitis virus G protein (VSV-G) pseudo-typed self-inactivating (SIN) [23, 24] HIV-1-based lentiviral vector pRRL-SIN-18-cPPT-hPGK-EGFP-WPRE [25] with a 118-bp sequence containing the central polypurine tract (cPPT) and central termination sequences as well as the woodchuck hepatitis post-transcriptional regulatory element (WPRE), which augments transgene expression. This was acquired from L Naldini (Milan, Italy) at a viral concentration of 1.7 × 109 transforming units per milliliter. This vector encloses the enhanced green fluorescent protein (eGFP) reporter gene driven by the human phosphoglycerate kinase gene (hPGK) promoter.

hfMSCs were plated at a density of 4 × 104 per cm2 in 24-well plates and a single round transduction in the presence of 2 μg/ml of polybrene (Sigma) for 12 hours. Parallel controls of hfMSCs were set up with only the addition of polybrene in D10. Transduced MSCs were harvested using trypsin EDTA 2 days after infection, washed twice with CellWash (Becton-Dickinson, Oxford, U.K.). Data were acquired on a FACS-calibur cell cytometer (Becton-Dickinson) and analyzed with CellQuest software (Becton-Dickinson). For analyzing the expression of the GFP reporter gene, cells were excited at 488 nm and detected in the FL1 channel at 530/30 nm.

Doubling Time of hfMSCs after Transduction

To determine effects of retroviral transduction on self-renewal, the doubling times of both transduced and mock-transduced cells (n = 3) was assessed until the cells senesced and stopped dividing. Transduced and mock-transduced hfMSCs were plated in triplicate at 5×105 per 100mm dish in D10, allowed to grow to subconfluence before being trypsinized, and counted in a hemocytometer. This was repeated until the cells reached senescence.

Differentiation Post-Transduction

Osteogenic and adipogenic differentiation were performed as previously described [15]. Briefly, transduced and mock-transduced hfMSCs were plated at 2 × 104 per cm2 in either osteogenic media (D10 supplemented with 10 mM β-glycerophosphate, 0.2 mM ascorbic acid, and 10−8 M dexamethasone; Sigma) or adipogenic media (D10 supplemented with 5 μg/ml insulin, 10−6 M dexamethasone, and 60 μM indomethacin; Sigma) for up to 4 weeks. Evidence of osteogenic differentiation was sought from Von Kossa staining of extracellular calcium crystals, which stain black. Adipogenic differentiation was evidenced by the appearance of lipid inclusion vacuoles within the cells, which take up the neutral lipid oil red O.

For myogenesis, hfMSCs were plated at 2 × 104 per cm2 in fibronectin-lined chamber slides and cultured in serum-free conditions in the presence of myoblast-conditioned media [26]. After 7 to 12 days in culture, evidence of multinucleated myotubes was sought by immunostaining for the muscle-specific protein desmin. Cells were fixed in methanol:acetone (1:1), blocked with nonserum protein block (X0909, Dako Cytomation, Cambridgeshire, U.K.) before incubation with the monoclonal mouse antidesmin (D33 clone, Dako Cytomation) at 1:200 dilution. This was followed by a biotinylated secondary goat anti-mouse antibody (BA-9200, Vector Laboratories Inc, Burlingame, CA). Cells were finally labeled with a streptavidin-conjugated fluorescein (SA-5001, Vector Laboratories Inc.) or alexa fluor 594 (S11223, Molecular Probes Europe BV, Leiden, The Netherlands). Slides were analyzed by epifluorescence microscopy (fluorescence microscope; Zeiss Axioskope, Jena, Germany), and images were captured using a cooled charge-coupled device camera and reviewed in Quipps m-FISH software (Vysis, Richmond, U.K.).

Transgene Expression Post-Transduction

Transgene expression was assessed at early (2 to 3 days after transduction) and late (14 weeks after transduction) time points. LacZ expression was analyzed by X-Gal staining [27] according to standard methodology. Cells expressing the transgene product β-galactosidase acquire insoluble blue coloration. Four random low-powered fields (×10) were chosen (82 – 580 cells each), and the proportion of cells stained blue via X-Gal was enumerated and compared with nonstaining cells. For analysis of eGFP expression, flow cytometry analysis (FACScalibur, Becton-Dickinson) was done at 48 hours after transduction, passage 16, and again at passage 24.

Clonal Expansion of Lentiviral-Transduced MSCs

Single-cell expansion of hfMSCs was done via limiting dilution. Thirty GFP-positive hfMSCs were plated in a 96-well plate in D10 and allowed to settle overnight. Single cells were identified by fluorescent microscopy the next day to ensure any subsequent following colony was derived from a single cell. Clonal populations were followed daily and were trypsinized when 60% – 80% confluent and expanded in successive passage in D10. Clonal populations were harvested when cells reached confluence in a 100mm plate, and tests of differentiation capacity were performed as described above.


Data are expressed as mean and standard deviation. Paired and nonpaired parametric data were compared by Student's t-test.A p value of < .05 was considered significant.


Transduction Efficiency

Human fetal mesenchymal stem cells were transduced initially with a onco-retroviral vector encoding the LacZ transgene (Fig. 1A). Static transduction of three samples of first-trimester fetal blood-derived hfMSCs from 9+1, 9+6, and 10+6 weeks of gestation (derived from three different donors) revealed a mean efficiency of 18.1 ± 8.6%, 43.0 ± 2.8%, and 45.1 ± 2.8% after 1, 2, and 3 cycles of transduction, respectively (Fig. 2A). Transduction under centrifugational forces demonstrated roughly a twofold increase in transduction efficiency (n = 3) from 39.0 ± 1.8% to 80.2 ± 5.4% after 2 cycles (paired t–test; p = .002) (Fig. 2B). Beyond two cycles, there was no observed increase in transduction efficiency with either static or centrifugational transduction.

Figure Figure 1..

(A): Schematic diagram of a Maloney murine leukemia retroviral construct encoding the LacZ reporter gene. (B): Schematic diagram of vesicular stomatitis virus G protein pseudo-typed self-inactivating HIV-1–based lentiviral vector. Abbreviations: cPPT, central polypurine tract; EGFP, enhanced green fluorescent protein; GA, 5′ portion of the gag gene included in the encapsidation signal (ψ); hPGK, human phosphor-glycerokinase; IRES, internal ribosome entry site; LTR, long terminal repeats; RRE, rev-response element; SIN, self-inactivating; Wpre, woodchuck hepatitis post-transcriptional regulatory element.

Figure Figure 2..

(A): Effect of transduction cycle number on efficiency of transduction. (B): Static transduction versus centrifugational transduction (mean, standard deviation). (C): X-Gal staining after retroviral transduction and G418 antibiotic selection, demonstrating up to 95% of cells expressing LacZ (×20). (D): Percentage of lentiviral-transduced cells with varying multiplicity of infection. (E): Mean fluorescence intensity with varying multiplicity of infection after lentiviral transduction. (F): GFP expression of hfMSCs after lentiviral transduction (×20). (G): FACS histogram of hfMSCs 48 hours after lentiviral transduction demonstrating 96% cells positive for GFP expression. (H): FACS histogram showing transduced hfMSCs after 14 weeks in culture with 95% of cells expressing GFP. (I): FACS histogram of control mock-transduced hfMSCs. Abbreviations: FACS, fluorescence-activated cell sorter; GFP, green florescent protein; hfMSC, human fetal mesenchymal stem cell.

Using a lentiviral vector, transduction efficiencies of 97.7 ± 1.4% with mean fluorescence intensities (MFIs) of 2,395 ± 460 were achieved with the same three samples. This was achieved after a single cycle at a MOI of 11. Raising viral titers was detrimental to the hfMSCs, with massive cell death observed after MOI of > 22 (Figs. 2D, 2E).

Growth Kinetics and Differentiation

After selection of the retroviral-transduced cells with the neomycin analogue Geneticin (G418) for 6 days (95% of cells now expressing LacZ as evaluated by X–Gal staining, Fig. 2C), the doubling time of both transduced and mock-transduced hfMSCs (n = 3) was recorded until the cells senesced. Doubling time was similar in both retroviral-transduced and control cells, with senescence occurring at about the same time at passages 26 through 29 (14 weeks after transduction, corresponding to 45 to 50 population doublings from passage 8, Fig. 4A). The doubling time of lentiviral-transduced cells was similar to mock-transduced control hfMSCs. Once again, self-renewal capacity was not affected, with the cells doubling 45 to 50 times before senescing after more than 14 weeks in culture (Fig. 4B).

Differentiation potential was tested at 2 weeks and 14 weeks after transduction (passages 8 and 20). Both adipogenic (Figs. 3A, 3B) and osteogenic (Figs. 3D, 3E) differentiation were unaffected at the two time points, whereas myogenic differentiation was possible up to passage 14 (6 weeks after transduction) in both transduced and mock-transduced cells (Fig. 3H). Similarly, lentiviral-transduced hfMSCs retained the ability to undergo both adipogenic (Fig. 3C) and osteogenic (Fig. 3F) differentiation at both time points (2 and 14 weeks after transduction), whereas myogenic differentiation was again lost after passage 14 (6 weeks after transduction) at the late time points (Fig. 3I).

Figure Figure 3..

(A–C): Adipogenic differentiation of retroviral-, lentiviral-, and mock-transduced hfM-SCs. Oil red O staining. (D–F): Osteogenic differentiation of retroviral-, lentiviral-, and mock-transduced hfMSCs. Von Kossa staining. (G–I): Myogenic differentiation of retroviral-, lentiviral-, and mock-transduced hfMSCs. Desmin staining in fluorescein in (G) and (H) and CY3 (red) in (I). 4′ ,6′-diamidino-2-phenylindole in blue. Abbreviation: hfMSC, human fetal mesenchymal stem cell.

Figure Figure 4..

(A): Doubling time of transduced (▪) versus mock-transduced control (♦) hfMSCs (mean, standard deviation). (B): Doubling time of lentiviral-transduced (▪) versus mock-transduced (♦) cells (error bars, standard deviation). Abbreviations: GFP, green fluorescent protein; hfMSC, human fetal mesenchymal stem cell.

Transgene Expression Over Time

Transgene expression in retroviral-transduced hfMSCs (percent of LacZ expression) decreased from 95% to 48.3 ± 3.9% (p = .03) over 14 weeks in culture (passage 24), whereas transgene expression in lentiviral-transduced hfM- SCs after 14 weeks in culture revealed no significant change in the percentage of cells expressing eGFP at 94.6 ± 2.6% (versus 97.7 ± 1.4%) and mean fluorescence intensity of 2607 ± 411 (versus 2395 ± 460) (Figs. 2G–2I). Fluorescence-activated cell sorter (FACS) analysis at passage 16 (5 weeks after transduction) showed similar results, with 94.5 ± 0.3% expressing eGFP with a MFI of 2843 ± 166 (not significant).

Clonal Expansion of Lentiviral-Transduced Cells

A total of four clones were derived after plating 30 cells in 96–well plates. After 49 days, 20.7 population doublings were achieved, giving a clonal efficiency of 13.3%. These clones were capable of both osteogenic and adipogenic differentiation, as in the nonclonal-transduced cell populations.


In demonstrating that hfMSCs can be efficiently transduced, we investigated several parameters essential for successful ex vivo gene therapy. First, we transduced hfM-SCs with both lentivirus and MoMLV vectors carrying reporter genes to address the possibility of gene expression shut down, which is known to occur with retroviral [2830] but not lentiviral [31, 32] vectors. Then we studied the effects of gene transfer and expression on hfMSC morphology, self-renewal capacity (over 14 weeks, 50 population doublings), and multipotentiality along adipogenic, osteogenic, and myogenic lineages.

High-efficiency gene transfer to hf MSCs with onco-retrovirus was achieved by combining the use of amphotropic retroviral particles derived from a second-generation 293T packaging cell line with a two-cycle centrifugal transduction procedure. The mechanism of centrifugation increasing transduction efficiency has not been elucidated, although other investigators have found similar results [5, 33]. As expected, the expression of the retroviral transgene declined substantially with time over 14 weeks in culture [28, 30].

The development of lentiviral vectors circumvents some of the short comings of onco-retroviral vectors, because they demonstrate efficient transduction of both quiescent and actively diving cells [31, 34] and long-term transgene expression both in vitro [31] and in vivo [32]. Lentiviral transduction of hfMSCs was significantly more efficient than with the MoMuLV vector with maintenance of transgene expression over 14 weeks in culture (45 to 50 population doublings).

The growth potential of hfMSCs, as assessed by their doubling time over 14 weeks in culture, was unaffected by either onco-retroviral or lentiviral transduction. hfM-SCs retained the potential for adipogenic and osteogenic differentiation at both early and late time points (corresponding to 2 and 14 weeks after transduction), although myogenic differentiation was lost after passage 16 in both transduced and mock-transduced cells for reasons that are not known at present. This is not attributable to the effects of transduction, because both transduced and control cells behaved in a similar manner.

These results confirm and extend our earlier pilot result using the pLX-based retroviral vector, where efficiencies of up to 98% were achieved with maintenance of transgene expression over a short time period (6 weeks). They also compare well with results from adult BM-derived MSCs. Lee et al. [5] reported retroviral transduction efficiencies of up to 80% with centrifugation and maintenance of trans-gene expression over 6 months in culture, although unlike our study, there was no comparison with mock-transduced controls [5]. Zhang et al. [31] also demonstrated lentiviral transduction efficiencies of up to 90% with maintenance of transgene expression over 5.5 months, but unlike our study, did not compare self-renewal capability with mock-transduced cells. Other groups reported varying transduction efficiencies, with Reyes et al. [35] reporting 30% to 70% retroviral transduction efficiencies with BM-derived multipotent adult progenitor cells and Morizono et al. [36] reporting up to 94% lentiviral transduction efficiency with adiposal-derived MSCs.

hfMSCs have a higher proliferative capacity [20] than adult BM-derived MSCs [37], which should facilitate clonal analysis of transduced cells. This will be important to ensure vector integration within safe genomic loci before transplantation in the wake of recent reports of T cell leukemia attributed to retroviral proto-oncogene transactivation [38]. We show here that clonally derived lentiviral-transduced cells can be expanded for at least 20 population doublings and still maintain differentiation capacity. This raises the possibility of additionally analyzing these cells for safe genomic loci.

Although early samples between 9 and 11 weeks were used for the current experiments, we have found no difference with gestational ages (7 through 14 weeks amenorrhea) in their basic immunophenotype, self-renewal, or differentiation ability [15] and thus expect results to be applicable across the late first trimester. High transduction efficiency and stable transgene expression are prerequisites for long-term therapeutic efficacy to overcome the low level of engraftment. Transduction levels in hfMSCs from fetuses at as early as 9 weeks of gestation are higher than shown with hemopoietic stem cells but comparable with adult BM-derived MSCs [5, 31]. Given the efficiency of transduction and the rapid expansion of hfMSCs, one could generate sufficient numbers of transduced cells to reinfuse within a few weeks of harvest for ex vivo gene therapy. Ultrasound-guided techniques have already developed to the extent that fetal blood sampling can be performed in ongoing pregnancies as early as 12 weeks with 5% loss rate [39], and advances in imaging and thin-gauge fetoscopy should render MSC harvest and intraperitoneal reinfusion feasible in the late first or early second trimester [40, 41].

To our knowledge, this is the first report demonstrating gene transfer into fetal blood-derived MSCs with maintenance of transgene expression, self-renewal, and differentiation capability at both short and long time points. We suggest that circulating hfMSCs may be efficacious targets for prenatal ex vivo gene therapy. Studies are now indicated to evaluate the effects of engraftment on and the stability of transgene expression in experimental models.


We acknowledge grant support from the Hammer-smith Hospitals Trust Research Committee and the Institute of Obstetrics and Gynaecology Trust. We thank L. Naldini (Milan) for the supply of the lentiviral vector, G. Nolan (Stanford, CA) for the supply of the retroviral vector and packaging cell line, K. Brimah and H. Kurata for their technical assistance, and M. Themis for his helpful suggestions.