A neovascularized organoid derived from retrovirally engineered bone marrow stroma leads to prolonged in vivo systemic delivery of erythropoietin in nonmyeloablated, immunocompetent mice


Marrow stromal cells (MSCs) are postnatal progenitor cells that can be easily cultured ex vivo to large amounts. This feature is attractive for cell therapy applications where genetically engineered MSCs could serve as an autologous cellular vehicle for the delivery of therapeutic proteins. The usefulness of MSCs in transgenic cell therapy will rely upon their potential to engraft in nonmyeloablated, immunocompetent recipients. Further, the ability to deliver MSCs subcutaneously – as opposed to intravenous or intraperitoneal infusions – would enhance safety by providing an easily accessible, and retrievable, artificial subcutaneous implant in a clinical setting. To test this hypothesis, MSCs were retrovirally engineered to secrete mouse erythropoietin (Epo) and their effect was ascertained in nonmyeloablated syngeneic mice. Epo-secreting MSCs when administered as ‘free’ cells by subcutaneous or intraperitoneal injection, at the same cell dose, led to a significant – yet temporary – hematocrit increase to over 70% for 55±13 days. In contrast, in mice implanted subcutaneously with Matrigel™-embedded MSCs, the hematocrit persisted at levels >80% for over 110 days in four of six mice (P<0.05 logrank). Moreover, Epo-secreting MSCs mixed in Matrigel elicited and directly participated in blood vessel formation de novo reflecting their mesenchymal plasticity. MSCs embedded in human-compatible bovine collagen matrix also led to a hematocrit >70% for 75±8.9 days. In conclusion, matrix-embedded MSCs will spontaneously form a neovascularized organoid that supports the release of a soluble plasma protein directly into the bloodstream for a sustained pharmacological effect in nonmyeloablated recipients.


In vivo delivery of replication-defective adenoviral (Ad)1,2,3,4,5 and adeno-associated viral (AAV) vectors6 encoding for therapeutic plasma proteins such as erythropoietin (Epo) has been described.7,8,9,10 However, the utilization of Ad vectors3,11 and of AAV vectors12,13 may be limited by their potential ability to elicit a lethal host immune response14,15,16,17 or possibly malignant complications.18 Nonviral approaches for Epo delivery have also been realized through naked plasmid DNA injection and gene electrotransfer.19,20,21,22,23 As compared to viral vectors, gene expression from plasmid DNA may be insufficient to provide therapeutic protein levels, especially in larger mammals.24,25 Although gene therapy strategies are promising, alternative strategies for sustained delivery of therapeutic plasma proteins are worth developing. Transgenic cell therapy, where autologous cells are genetically engineered into synthetic endocrine cells and subsequently returned to the donor, shows promise and has been tested in clinical trials.26 A crucial component to this strategy will be to choose an autologous cellular source of sufficient expansion potential to allow for a sizable clonal implant. Indeed, bone marrow stromal cells (MSCs), which play a crucial role in the bone marrow microenvironment and in hematopoiesis, may constitute an ideal autologous cellular vehicle for the delivery of therapeutic gene products.27,28,29,30,31 MSCs are present in sufficient numbers in humans of all ages, they can be harvested by bone marrow aspiration in the absence of prior mobilization treatment, and they can differentiate into many mesenchymal phenotypes such as osteoblasts, chondrocytes, adipocytes, fibroblasts, skeletal myoblasts, cardiomyocytes, tenocytes, and smooth muscle cells.29,30,32,33 Studies have shown that MSCs maintain their precursor phenotype following gene modification and culture expansion, and are therefore promising for cell and gene therapy applications.30 In their native ‘pluripotent’ state, MSCs may be utilized clinically for the treatment of mesenchymal disorders, such as osteogenesis imperfecta.34,35 The strong proliferative capacity of MSCs renders these amenable to retroviral gene transfer,29 and subsequent expansion for utilization as a vehicle for soluble protein secretion.36 The use of MSCs has been demonstrated for the expression of an exogenous gene product in vitro and as an efficient cellular vehicle for the release of proteins in vivo.29,37,38,39,40,41 Various studies have shown the delivery of a secretable human transgene product by MSCs administered via the intravenous route in immunodeficient rodents.40,42,43 When infused intravenously in immunocompetent recipients, genetically engineered MSCs engendered transiently detectable plasma levels of a species mismatched secreted protein.44,45 Therefore, the successful utilization of MSCs in transgenic adoptive cell therapy will depend upon their demonstrated ability to engraft in nonmyeloablated, immunocompetent recipients and lead to a sustained pharmacological effect. We have previously shown in a pilot study that intraperitoneal implantation of gene-modified MSCs leads to pharmacological production of a plasma protein in unconditioned normal hosts.41 However, intravenous or intraperitoneal delivery of large amounts of engineered cells will not allow for their subsequent removal if unforeseen complications were to arise from their use. Therefore, we propose that as an alternative delivery route, engineered MSCs can be returned as a subcutaneous implant.

To test this hypothesis and as a proof-of-concept, we have utilized MSCs retrovirally engineered to secrete Epo. Long-term pharmacological effect can be monitored unambiguously by serial measurement of hematocrit (Hct) in experimental animals and will allow for comparison of various delivery routes. We report that the levels of mEpo released in vivo from MSCs implanted subcutaneously or in the abdominal cavity are sufficient to cause a supraphysiological effect as evidenced by a significant increase of blood Hct. Moreover, Epo-secreting MSCs mixed in Matrigel were shown to participate in de novo blood vessel formation and to differentiate into CD31+ cells. These data demonstrate that MSCs have the plasticity to spontaneously form a neovascularized organoid and that this platform will allow for sustained delivery of pharmacological levels of soluble plasma proteins such as Epo.


Titer of retrovirus producers

To determine gene transfer efficiency and transgene expression in stably transfected retroviral producer cells, flow cytometry analysis for green fluorescent protein (GFP) expression was performed. The proportion of GFP-positive cells in the polyclonal producer populations GP+E86-Epo-IRES-EGFP and Sorted GP+E86-Epo-IRES-EGFP, based on green fluorescence, was 34 and 97%, respectively. To evaluate the quantity of infectious particles released by these producers, a titration assay using their retroviral supernatant was conducted and the viral titers obtained were 2.4×105 and 4.0×105 infection particles/ml, respectively. The percentage of LacZ-positive cells in the GP+E86-LacZ viral producer cell population was >95% and the viral titer of these cells was 1.1×105 infectious particles/ml.

Retrovector expression and mEpo secretion by gene-modified marrow stroma

To determine the molecular genetic stability of the Epo-IRES-EGFP retroviral construct (Figure 1), proviral DNA in the genome of polyclonal retrovirally transduced MSCs was analyzed by Southern blot. A probe complementary to the GFP reporter allowed the detection of a DNA band consistent with the 3436 bp fragment anticipated from EcoRV digest of integrated unrearranged Epo-IRES-EGFP proviral DNA (Figure 2). No subgenomic or rearranged retrovector integrant was detected.

Figure 1

Schematic illustration of retroviral plasmid construct Epo-IRES-EGFP. The retroviral vector Epo-IRES-EGFP was generated as described in the Materials and methods section. The murine erythropoietin (Epo) cDNA sequence was inserted in the multiple cloning site of our previously described pIRES-EGFP construct, upstream of an internal ribosomal entry site (IRES) and the EGFP reporter gene. The retrovector pIRES-EGFP served as the control plasmid.

Figure 2

Southern blot analysis of Epo-IRES-EGFP-transduced mouse marrow stroma. Genomic DNA from Epo-IRES-EGFP-transduced primary murine MSCs was digested with EcoRV, which cuts once in each flanking LTR, and fractionated on 1% agarose gel. Hybridization of the blot with a 32P-labeled GFP probe permits detection of integrated, proviral DNA of predicted 3436 bp size. Untransduced murine MSCs served as the negative control (lane -). Molecular weights are indicated.

Retrovector expression in genetically engineered murine MSCs was confirmed by flow cytrometry analysis for GFP expression. The proportion of Epo-IRES-EGFP-modified MSCs expressing GFP was 91%. To establish that murine MSCs transduced with Epo-IRES-EGFP secrete mEpo in vitro, and quantitate the level, supernatant collected from these cells was analyzed by ELISA for human Epo. The Epo-IRES-EGFP-modified MSC population was analyzed and found to secrete 17 U of Epo per 106 cells per 24 h. The percentage of LacZ-positive cells in the LacZ-Epo-IRES-EGFP-modified MSC population was >90%. LacZ-Epo-IRES-EGFP-modified stroma was noted to secrete 17 U of Epo per 106 cells per 24 h. There was no Epo detected in the supernatant collected from control IRES-EGFP-transduced MSCs and LacZ-IRES-EGFP MSCs (data not shown).

Intraperitoneal implantation of Epo-secreting MSCs

We determined if mEpo secretion from Epo-IRES-EGFP-transduced MSCs implanted by intraperitoneal injection in nonmyeloablated, immunocompetent mice can lead to a measurable effect on Hct. We also established if there is a dose–response relationship between the number of Epo-IRES-EGFP-modified stromal cells injected and the resulting Hct. Cohorts of mice were implanted with either 105, 106, 5×106, or 107 Epo-IRES-EGFP-engineered MSCs. Peripheral blood was collected, and Hct and plasma Epo concentration measured over time as shown in Figure 3. As illustrated in Figure 3a, the Hct of mice that received 105 mEpo-secreting stromal cells rose to a peak value of 60±1.1% at 5 weeks postimplantation. In mice injected with 106 Epo-IRES-EGFP-transduced MSCs, blood Hct rose to a maximum of 68±3.8% at 2 weeks following implantation and then quickly declined to a steady 61% observed until week 12. The recipients of 5×106 mEpo-secreting MSCs had an increase in Hct that attained a value of 78% at 2 weeks postimplantation, remaining above 75% until 7 weeks following stroma administration. The Hct of mice implanted with 107 of these gene-modified MSCs attained the highest level at 4 weeks (88%), thenceforth persisting at 85% or greater up to week 9 and over 70% up to week 12. A parallel group of mice received 107 IRES-EGFP-transduced MSCs. These control mice maintained Hct levels ranging between 51 and 57% throughout this study (Figure 3a). A tight correlation was revealed between the number of i.p. implanted Epo-secreting MSCs and the resulting peak in the Hct (r=0.97).

Figure 3

Hct and plasma Epo concentration of mice following intraperitoneal implantation with mEpo-secreting marrow stromal cells. Mice were injected i.p. with a range of mEpo gene-modified MSCs. Epo-IRES-EGFP-transduced MSCs prior to implantation secreted 17 U of Epo per 106 cells per 24 h. Peripheral blood from the saphenous vein was collected over an 18-week period, and Hct (a) as well as plasma Epo (b) levels were assessed. Control mice were injected i.p. with IRES-EGFP gene-modified MSCs. Mean (n=3–5)±s.e.m. Error bars smaller than icons do not appear.

To quantify the plasma concentration of mouse Epo in mice administered Epo-IRES-EGFP-engineered marrow stroma, plasma Epo levels were measured by human Epo ELISA. As done by others in the field,5,49 we utilized ELISA kits for the detection of human Epo to detect mouse Epo. Although affinity for mEpo is poor,50 it remains the standard in the field and serves as a basis for comparison. Therefore, our measured plasma mEpo concentrations are likely underestimated because of levels below the threshold of detectability of this kit. Mice that received by intraperitoneal injection 107 and 5×106 Epo-IRES-EGFP-engineered MSCs secreting in vitro 17 U of Epo per 106 cells per 24 h, exhibited a rise in plasma Epo levels to 740±20 and 298±25 mU/ml, respectively, at 3 days postimplantation (Figure 3b), which declined proportionally by over 50% to 333±60 and 141±15 mU/ml, respectively, at 1 week, and by over 65% to 255±15 and 96±18 mU/ml, respectively, at 2 weeks. The concentration of Epo detected in the plasma of these mice at 7 weeks or greater postimplantation was under 20 mU/ml.

Subcutaneous implantation of Matrigel-embedded Epo-secreting MSCs

As an alternative delivery route, we tested whether mEpo-engineered MSCs implanted in the subcutaneous space display the same pharmacological features as intraperitoneal delivery. We conducted subcutaneous implantations of gene-modified MSCs premixed in Matrigel. Peripheral blood was collected, and Hct and plasma Epo concentration measured over time as represented in Figure 4. To first ascertain if there is a correlation between the number of Epo-secreting MSCs mixed in Matrigel and the consequent rise in Hct during the first 4 weeks postimplantation, groups of C57Bl/6 mice were injected subcutaneously with 4, 0.5, and 0.25×106 LacZ-Epo-IRES-EGFP-modified MSCs per mouse. The Hct of these mice increased from a baseline of 53±3% (mean±s.d.) to 90±1, 76±2, and 67±1%, respectively, within 2 weeks following implantation, as shown in Figure 4a. The Epo-secreting MSC dose and the resulting Hct correlated strongly (r=0.90). The Hct of the negative control group generated by implantation of Matrigel-embedded LacZ-IRES-EGFP MSCs maintained the baseline values (51±3%) over the 4-week period of the experiment. As a comparison, we determined the effect of Matrigel admixed with recombinant human Epo (rhuEpo) only. We found that in mice implanted with Matrigel/rhuEpo (1000 U in 0.5 ml of Matrigel or 40 000 U/kg), the Hct increased from 50±2 to 63±2% within 2 weeks and was thereafter sustained for the subsequent 2 weeks. The pattern in the change of Hct over time with rhuEpo was similar to that achieved when mice received the lowest tested dose of 0.25×106 Epo-secreting MSCs (Figure 4a).

Figure 4

Hct and plasma Epo concentration of mice following subcutaneous implantation with mEpo-secreting marrow stromal cells embedded in Matrigel™. Mice were injected s.c. with different numbers of mEpo gene-modified MSCs embedded in Matrigel. Epo-IRES-EGFP-transduced MSCs prior to implantation secreted 17 U of Epo per 106 cells per 24 h. Peripheral blood from the saphenous vein was collected over a 4-week period, and Hct (a) as well as plasma Epo (b) levels were determined. Mice were also implanted with Matrigel mixed with 1000 U of Eprex, and control mice received IRES-EGFP gene-modified MSCs embedded in Matrigel. Mean (n=3–5)±s.d. Error bars smaller than icons do not appear.

To determine the concentration of mouse Epo in blood plasma of mice subcutaneously injected with Epo-secreting MSCs embedded in Matrigel, human Epo ELISA was performed. In mice implanted with these Matrigel-embedded MSCs, the plasma Epo concentration increased from <30 mU/ml prior to implantation to 510, 280, and 270 mU/ml with 0.25×106 LacZ-Epo-IRES-EGFP-modified MSCs at 1, 2, and 3 weeks postimplantation, respectively (Figure 4b). At these time points, 0.5×106 LacZ-Epo-IRES-EGFP MSCs led to plasma Epo levels of 700, 540, and 570 mU/ml. Values observed at 4 weeks were similar to those at 2 and 3 weeks following implantation. In mice implanted with Matrigel mixed with LacZ-IRES-EGFP MSCs or rhuEpo, the concentration of Epo detected was <35 mU/ml. Unlike the change in Hct observed over time with rhuEpo, plasma Epo levels were not altered (Figure 4), which is likely because of the short half-life of the recombinant protein (24 h when administered by subcutaneous injection) and the nonsustained nature of its delivery as opposed to continuous Epo release achieved when Epo-secreting MSCs are embedded in Matrigel.

In vivo endothelial differentiation of Matrigel-embedded Epo-secreting MSCs

To study the in vivo fate of Epo-secreting MSCs mixed in Matrigel, these cells were gene modified to also express β-galactosidase (LacZ-Epo-IRES-EGFP MSCs). X-gal histochemical analysis of surgically excised implants was subsequently performed at 4 weeks postimplantation. Macroscopic examination revealed the occurrence of blood vessels within MSC-containing Matrigel implants (Figure 5a). Sections of the implant were prepared to show transgene-expressing cells based on LacZ gene reporter activity. By X-gal staining, we detected the β-galactosidase-expressing Epo-producing MSCs randomly dispersed within the implant with a fibroblast-like histological appearance but additionally, as shown in Figure 5b, incorporated in the wall of blood vessels. As evidenced in Figure 5c, these cells had adopted the histological configuration of endothelial cells and had become CD31+, results consistent with the in vivo phenotypic differentiation of MSCs into endothelium.

Figure 5

In vivo differentiation of Matrigel-embedded Epo-secreting marrow stromal cells into CD31+ endothelial cells. Epo-IRES-EGFP-transduced MSCs subsequently gene modified with LacZ retroparticles were mixed with Matrigel, and implanted by subcutaneous injection in normal, unconditioned C57Bl/6 mice, at a density of 4×106 cells in 500 μl Matrigel. Four weeks later, recipient syngeneic mice were killed, implants retrieved, and sections prepared and X-gal stained to show β-galactosidase-expressing donor stroma. Macroscopically evident neoangiogenesis occurred in MSC-containing implants (a). Representative hematoxylin and eosin stained sections from Matrigel implants comprising Epo-secreting MSCs show a cross-sectioned blood vessel containing LacZ+ marrow stromal cells (green-blue color) (b, ×40 magnification), some of which are immunohistochemically stained with anti-mouse CD31 antibodies (light-brown color) (c, ×100 magnification).

Long-term Hct following subcutaneous implantation of Epo-secreting MSCs in Matrices

In order to assess if providing MSCs with an artificial microenvironment is of importance for sustained pharmacological production of Epo, we compared the long-term impact on Hct of MSCs delivered freely in the subcutaneous space with MSCs mixed in Matrigel. As shown in Figure 6a, in C57Bl/6 mice implanted with 4×106 Matrigel-embedded Epo-IRES-EGFP MSCs, the Hct increased from a basal 55±0.7% (mean±s.e.m.) to 82±1.2% at 17 days postimplantation and persisted at levels of 80–90% until day 70 in one mouse, and for over 300 days in the other two recipient mice. Control mice were generated by implantation with 4×106 Matrigel-embedded IRES-EGFP MSCs and demonstrated a consistent Hct of 55% over time (Figure 6a). In a separate experiment where 4×106 LacZ-Epo-IRES-EGFP MSCs mixed in Matrigel were injected in another three mice, two of three recipient animals showed Hcts above 80% from day 22 to past day 118 postimplantation (Figure 6a). Control mice (n=3), which received 4×106 LacZ-IRES-EGFP MSCs in Matrigel, maintained an Hct of 55% (Figure 6a). In contrast, for the same number of Epo-IRES-EGFP MSCs in the absence of Matrigel, the Hct rose from a basal 56±0.3% (mean±s.e.m.) before implantation to a peak level of 85±0.9% at 14 days postimplantation which persisted for an additional 14 days, and thereafter declined rapidly in four of five mice and attained basal values at 50 days (Figure 6b). One mouse maintained Hct values above 70% for 150 days. Control mice implanted with 4×106 MSCs engineered with an Epo-less retrovector demonstrated stable Hct levels of 55% (Figure 6b). A significant difference on long-term effect on Hct was noted between the Matrigel-embedded Epo-secreting MSCs when compared to the unembedded cells (P=0.0348 LogRank).

Figure 6

Long-term Hct of mice following subcutaneous implantation of mEpo-secreting marrow stromal cells with or without Matrigel. Epo-secreting MSCs either embedded in Matrigel (a) or without a matrix (as ‘free cells’) (b) were injected s.c. in C57Bl/6 syngeneic mice (n=3–5). Epo-IRES-EGFP-transduced MSCs (full square) as well as LacZ-Epo-IRES-EGFP-transduced MSCs (full triangle) prior to implantation secreted 17 U of Epo per 106 cells per 24 h. Peripheral blood from the saphenous vein was collected for up to a maximum 320-day period, and Hct ascertained. Control mice received IRES-EGFP gene-modified MSCs (as indicated by the x icon in the figure) and/or LacZ-IRES-EGFP MSCs (+icon).

Matrigel is immunologically incompatible with nonmurine species. Among the many components of Matrigel, collagen figures prominently and may play an important role as part of the artificial microenvironment provided by Matrigel to MSCs. We hypothesized that a human-compatible type I bovine-derived collagen pharmaceutical-grade product could serve as a substitute for Matrigel, thereby offering clues toward clinically feasible application of this strategy. As shown in Figure 7, 4–5×106 collagen-embedded Epo-IRES-EGFP MSCs led to a significant increase in Hct compared with controls (Figure 7). Specifically, in mice (n=5) implanted with Collagen-Matrix-embedded Epo-secreting MSCs, the Hct increased from 55±0.3% to a peak level of 82±2.4% at 20 days and thereafter gradually decreased. A significant difference in Hct was observed between mice implanted with Collagen-Matrix-embedded Epo-secreting MSCs and control mice (P<0.001) (Figure 7). The effect on Hct was lost in all mice by 120 days postimplantation. We noted that the decline in Hct was concurrent with the physical disappearance of the implant that was palpable in the first weeks and gradually resorbed. When comparing the long-term effect on Hct, all mice implanted with Collagen-Matrix-embedded Epo-secreting MSCs sustained an Hct above 70% for 75±8.9 days, whereas in most mice (four of five) that received unembedded cells, this level lasted 32±1.5 days.

Figure 7

Long-term Hct of mice following subcutaneous implantation of mEpo-secreting marrow stromal cells embedded in a human biocompatible type I bovine collagen matrix. Epo-secreting MSCs embedded in Collagen Matrix were implanted s.c. in C57Bl/6 syngeneic mice. Epo-IRES-EGFP transduced MSCs prior to implantation secreted 17 U of Epo per 106 cells per 24 h. Peripheral blood samples from the saphenous vein were harvested for up to 160 days, and the Hct determined. One group of test mice (n=5) received these Epo-IRES-EGFP gene-modified MSCs in Collagen Matrix while two groups of control mice received the IRES-EGFP transduced MSCs in Collagen Matrix (n=5) or the Collagen Matrix alone (n=5).


A variety of autologous cellular vehicles for delivery of Epo have been explored, including skeletal myoblasts49,51 and stomach-implanted vascular smooth muscle cells.52 Delivery of transgenes other than Epo has also been explored with hematopoietic stem cells, lymphocytes, fibroblasts, human umbilical vein endothelial cells, and blood outgrowth endothelial cells (BOECs).53 In contrast to MSCs, there are drawbacks associated with the use of most of these cells in an autologous setting. Skin fibroblasts have been shown to inactivate introduced vector sequences following transplantation54,55 and, depending on the age of the donor, may have limited in vitro proliferation capacity;56 this may limit the amount of engineered fibroblasts available for clinical use.26 Skeletal myoblasts are present in very low amounts in the majority of adult mammals, and their successful growth and transplantation is technically challenging.57,58,59 Vascular smooth muscle cells, to engraft in humans, may necessitate arterial injury.60,61,62 Hematopoietic stem cells can be difficult to expand in culture and gene-modify,63,64 and very large numbers are required for engraftment in the absence of a toxic ‘conditioning’ regimen.65 Lymphocytes possess a short lifespan,66 and human umbilical vein endothelial cells cannot be obtained from an adult.

In contrast to terminally differentiated cell types, autologous MSCs have a very robust expansion capacity and inherent plasticity that can be exploited therapeutically.30 Indeed, the safety of culture-expanded MSCs was validated in a clinical trial where autologous human MSCs were coinjected with peripheral blood stem cells in breast cancer patients undergoing chemotherapy.67 Similarly, no side effects occurred in patients of a phase I trial following the intravenous administration of up to 50 million cultured autologous MSCs.68 These data buttress the clinical feasibility of using autologous MSCs following a ‘conditioning’ regimen. Although practical in the setting of cancer therapy, toxic ‘conditioning’ preparatory regimens would be intolerable for cell therapy of nonmalignant disease. Hence, the practical use of MSCs for the secretion of therapeutic gene products in cell and gene therapy applications will rely on their capacity to engraft and persist in vivo following reimplantation,69 in the absence of ‘conditioning’ regimens. Several investigators have examined recombinant protein delivery by genetically engineered MSCs following their intravenous administration in immunodeficient hosts. Engraftment and persistence of human MSCs expressing human factor VIII, human factor IX, or human IL-3 was demonstrated in unconditioned immunodeficient mice following transplantation.40,42,43,70,71 Similar experiments with human transgene products have been carried out in immunocompetent transplant models. For example, gene-modified canine MSCs expressing human growth hormone or human factor IX were infused in unconditioned normal dogs. However, the human xenoprotein was only very transiently detected in the plasma39,44 and this was shown to be because of the production of neutralizing antibodies (canine anti-hFIX).39 Similarly, in baboons surgically implanted with human Epo gene-modified baboon MSCs placed in an immunoisolatory device, the detection of the human protein lasted up to 9–139 days.45

Although it is speculated that genetically engineered autologous MSCs will be useful for transgenic cell therapy of disease without requirement of prior myeloablation,69 the biological proof of this hypothesis still needs to be demonstrated in animals with intact immune and hematological systems. In an effort to address this issue, we gene-modified murine MSCs to produce murine Epo and tested their long-term effect on erythropoiesis in normal mice. We chose Epo as a ‘reporter’ transgene since its long-term in vivo biological activity is readily and objectively assessed by serial measurement of Hct over time. This experimental approach also serves as proof-of-principle for the utility of MSCs as cellular vehicles for pharmacological delivery of therapeutic plasma-soluble proteins.

We first generated a bicistronic retrovector comprising the murine Epo cDNA and the GFP reporter gene, transduced murine MSCs and confirmed mEpo transgene expression by ELISA. We then implanted the genetically engineered MSCs into unconditioned normal syngeneic mice subcutaneously or intraperitoneally. We noted a robust and sustained elevation of blood Hct, with the best long-term effect noted in subcutaneously implanted matrix-embedded cells. Ding et al27 have shown that intravenous infusion of autologous MSCs in normal mice will lead to their dispersion in spleen, lung, kidney, and marrow. In contrast, intraperitoneal injection of isogenic MSCs will not lead to systemic dispersion unless accompanied by sublethal irradiation.72 Several studies have indicated that in the absence of conditioning treatment or mesenchymal tissue defect in the host, systemic engraftment of MSCs, outside the implantation site, is unlikely to occur.73 This observation has been validated in allogeneic MSC transplantation in children with osteogenesis imperfecta.34 Based on these reports, we speculate that the MSCs described in this study delivered by intraperitoneal injection are unlikely to have migrated outside of the abdominal cavity to other organ sites. However, for clinical applications, if genetically engineered MSCs were to relocate to undesirable organ compartments or if serious side effects were to arise, it would be impossible to remove these cells. Indeed, in a published cell therapy trial where transfected fibroblasts were used, a very significant effort was deployed to isolate engineered cellular clones that remain untransformed by the ex vivo DNA transfection procedure.26 Therefore, the theoretical risk of malignant transformation – which arises from the use of integrating gene delivery platforms – is a strong incentive to explore implantation strategies that avoid widespread biodistribution of engineered cells, as would occur after intravenous or intraperitoneal implantation.

Cells sequestered within a subcutaneous implant would enhance safety for clinical trials, and for this reason we have explored the use of engineered MSCs embedded within a matrix. We speculate that the sustained therapeutic effect we observed utilizing Matrigel-embedded gene-modified MSCs is because of the release of soluble plasma protein Epo directly into the bloodstream. This is supported by our demonstration that the Epo-secreting MSCs mixed in Matrigel also elicited and participated in de novo blood vessel formation leading to a neovascularized MSC implant (Figure 5). Specifically, we showed the occurrence of these cells near and within vascular structures as well as their transdifferentiation into CD31+ cells. We have also noted that MSC-released VEGF likely plays a central role in the observed transdifferentiation of MSCs into endothelial lineage (data not shown). In our model, we found that a cell dose of approximately 160×106 cells/kg (or 4×106 cells per 25 g mouse) led to supraphysiological production of Epo with sustained Hct above 70%. In an analogous cell therapy approach, Naffakh et al74 achieved continuous secretion of Epo in mice following intraperitoneal implantation of neo-organs containing gene-modified skin fibroblasts. The utilization of 20 million retrovirally engineered fibroblasts secreting 17 U of Epo per 106 cells per 24 h in vitro led to an increase in Hct attaining a peak value of 82% at 8 weeks which persisted for 6 months following implantation. However, the limitations in expanding transfected clonal human fibroblast colonies derived from adult volunteers to sufficient numbers to obtain a sizable implant will always be a limitation to their use as a cellular delivery vehicle.26 Therefore, we speculate that human MSCs secreting comparable amounts of hEpo (or any other therapeutic plasma-soluble protein) may have a similar pharmacological effect, and that the cell dose required for an average 70 kg adult may be clinically realizable. Indeed, human postnatal MSCs have been reported to be able to divide more than 50 times. This expansion potential should easily allow for genetic engineering and clonal selection, and subsequent expansion to cell doses high enough for a pharmacological effect.

As was similarly postulated in an in vivo study examining systemic Epo levels following direct intramuscular delivery of adenovector,5 there may be various nonexclusive explanations for the drop in Hct observed in some mice utilizing gene-modified MSCs. There may be a decline in the amount of implanted MSCs over time because of suboptimal in vivo microenvironment, impeding their proliferation or survival, as suggested with other cells.74 Another reason for the eventual loss of recombinant Epo in plasma is in vivo inactivation of transgene expression or promoter shut-off in genetically engineered cells, as previously seen in other gene therapy approaches.54,55 Lastly, the immune system may eliminate transgene-expressing cells. It must be restated that our engineered cells expressed mEpo as well as the xenogenic GFP reporter. In our study, the GFP cDNA served as a genetic marker, reporter, and selectable agent for engineered cells concurrently expressing the nonselectable, linked transgene, mEpo cDNA. In primary bone marrow stromal cells, GFP allowed the evaluation of therapeutic gene transfer efficiency by fluorescence microscopy and flow cytometry analysis. Recent investigations clearly demonstrate that cytotoxic immune response may occur in immunocompetent animals against GFP-expressing syngeneic cells75,76.

Although Matrigel serves as a useful support vehicle for MSCs in mice, its composition derived from a murine sarcoma cell line precludes its use in humans owing to obvious bioincompatibility. Matrigel is composed of a very diverse admixture of proteins, growth factors, and basal membrane components including collagen. We speculated that collagen might be an adequate minimal component in mediating MSC survival in implants since integrin-mediated docking to extracellular matrix components has been reported to inhibit apoptosis in adherent cells.77 For this reason, we tested a human-biocompatible, bovine type I collagen porous formulation as an MSC delivery platform/artificial microenvironment in normal mice. Although a long-term effect on Hct was less sustained than that observed with Matrigel-embedded cells, our results suggest that type I bovine collagen may serve as an adequate minimal component and that addition of other selected basal membrane components – as found in Matrigel – may allow for development of a high-performance human-compatible matrix platform for MSCs. Human blood outgrowth endothelial cells (BOECs) share similar features to MSCs with regard to desirability as an autologous cellular vehicle for therapeutic proteins such as Factor VIII. Although they have been studied following intravenous administration in immunodeficient mice,53 their delivery within a collagen-based matrix may theoretically yield similar results as we have observed. Another promising autologous cell type, the recently reported multipotent adult progenitor cells or MAPCs, copurifying with mesenchymal stem cells from the bone marrow, may serve the purpose studied here as well.78

Interestingly, mice tolerated high Hcts well. Although elevated Hct may lead to premature death from stroke in humans, we did not observe this phenomenon in our test mice. Further, mice with high Hcts did not display behavioral or physiological signs of distress such as weight loss. A reason for the survival of mice with Hct values approaching 90% is perhaps due to the role of nitric oxide.79 We also showed that the magnitude of the augmentation in blood Hct caused by Epo release in vivo was dependent on the number of implanted gene-modified MSCs and on their ex vivo Epo secretion levels. Therefore, our study revealed that we could dose the therapeutic effect and obtain the desired effect by adjusting the amount of genetically engineered MSCs implanted. Alternatively, MSCs engineered with regulated vector systems could be utilized.37 Lastly, we speculate that engineered MSCs bereft of immunogenic reporter proteins – such as GFP or drug selectable markers like Neor – should also allow for a more sustained transgene production. Alternative human biocompatible selectable markers would be of use in this setting.80

In conclusion, our data validate the feasibility of utilizing gene-modified autologous bone marrow stroma as a vehicle for sustained systemic production of recombinant therapeutic proteins in immunocompetent recipients, without the major drawback of myeloablation. Furthermore, when delivered subcutaneously in a matrix, MSCs will spontaneously generate a neovascularized organoid. This report provides a basis for future applications of Matrix-embedded MSCs as a safe and efficient delivery vehicle of plasma soluble gene products in a sustained fashion in the treatment of a large spectrum of inherited or acquired illnesses.

Materials and methods

Cell culture of murine fibroblasts

GP+E86 ecotropic retrovirus-packaging cell line46 from American Type Culture Collection (ATCC) was cultured in Dulbecco's modified essential medium (DMEM) (Wisent Technologies, St Bruno, QC) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (Wisent) and 50 U/ml penicillin, 50 μg/ml streptomycin (Pen/Step) (Wisent). National Institutes of Health (NIH) 3T3 mouse fibroblast cell line, obtained from ATCC, was grown in DMEM with 10% FBS and 50 U/ml Pen/Step. All cells were maintained in a humidified incubator at 37°C with 5% CO2.

Generation of retroviral vector and of virus-producing cells

The retroviral plasmid vector pIRES-EGFP was previously generated in our laboratory.47 This construct comprises a multiple cloning site linked by an internal ribosomal entry site (IRES) to the enhanced green fluorescent protein (EGFP) (Clontech Laboratories, Palo Alto, CA, USA). The retroviral vector pEpo-IRES-EGFP (Figure 1) was synthesized by obtaining the cDNA for mouse Epo by BamH1 digest of a pBluescript-based construct graciously provided by Jean M Heard (Institut Pasteur, Paris) and ligating it with a Bam H1 digest of pIRES-EGFP.

For the manufacture of recombinant virus-producing cells, the pEpo-IRES-EGFP construct (5 μg) was linearized by Fsp1 digest and cotransfected, utilizing lipofectamine reagent (GIBCO-BRL, Gaithesburg, MD, USA), with 0.5 μg pJ6ΩBleo drug resistance plasmid48 generously given by Richard C Mulligan (Children's Hospital, MA, USA), into GP+E86 packaging cells. Stable transfectants were selected by a 5-week exposure to 100 μg/ml zeocin (Invitrogen, San Diego, CA, USA), thus giving rise to the polyclonal virus-producing GP+E86-Epo-IRES-EGFP cells. GFP expression in cells was assessed by flow cytometry analysis utilizing an Epics XL/MCL Coulter analyzer and gating viable cells based on the FSC/SSC profile. A population of Sorted GP+E86-Epo-IRES-EGFP producers was obtained following sorting of GP+E86-Epo-IRES-EGFP cells based on green fluorescence using a Becton Dickinson FACSTAR sorter. The control GP+E86-IRES-EGFP producers were generated in the same manner. Retroparticles from all producers were devoid of replication-competent retrovirus as was determined by the GFP marker rescue assay employing conditioned supernatants from transduced target cells. GP+E86-LacZ retrovirus producing cells were generated by transinfection of the GP+E86 cell line with filtered retroviral supernatant from 293GPG-LacZ producers (generously provided by RC Mulligan, Children's Hospital, MA, USA) twice per day for three consecutive days, in the presence of 6 μg/ml lipofectamine.

Titer determination of retrovirus producers

To assess the titer of GP+E86-Epo-IRES-EGFP and GP+E86-IRES-EGFP producers, NIH 3T3 fibroblasts were seeded at a density of 2–4×104 cells/well of six-well tissue culture plates. The next day, cells were exposed to serial dilutions (0.01–100 μl) of 0.45 μm filtered retroviral supernatants, in a total volume of 1 ml complete media with 6 μg/ml lipofectamine. Cells from extra test wells were counted and averaged to disclose the baseline cell number at the moment of virus addition. Three days later, the percentage of GFP-expressing cells was ascertained by flow cytometry analysis. The titer was calculated using the following equation by considering the virus dilution that yielded 10–40% GFP-positive cells. Titer (infectious particles/ml)=(% GFP-positive cells)×(amount of target cells at start of virus exposure)/(volume of virus in the 1 ml applied to cells). The titer of GP+E86-LacZ virus producers was estimated through X-gal staining of likewise transduced NIH 3T3 cells.

Harvest, culture, and transduction of murine bone marrow stroma

Whole bone marrow was harvested from the femurs and tibias of 18–22 g female C57Bl/6 mice (Charles River, Laprairie Co., QC, Canada) and plated in DMEM supplemented with 10% FBS and 50 U/ml Pen/Step. After 4–5 days of incubation at 37°C with 5% CO2, the nonadherent hematopoietic cells were discarded and the adherent MSCs were gene modified as follows. Media was removed from MSCs and replaced with 0.45 μm-filtered retroviral supernatant from subconfluent Sorted GP+E86-Epo-IRES-EGFP or control GP+E86-IRES-EGFP producers once per day for six consecutive days, for each of two successive weeks, in the presence of 6 μg/ml lipofectamine. The resulting genetically engineered stromal cells were subsequently expanded for 2–3 months. As additional populations of gene-modified MSCs, Epo-IRES-EGFP-modified MSCs as well as control IRES-EGFP MSCs were also transduced with retroparticles from GP+E86-LacZ producers twice per day for three consecutive days with 6 μg/ml lipofectamine, giving rise to LacZ-Epo-IRES-EGFP-modified MSCs and LacZ-IRES-EGFP MSCs, respectively. GFP expression in genetically engineered stroma was evaluated by flow cytometry analysis to allow an estimate of the gene transfer efficiency. Beta-galactosidase expression in LacZ gene modified MSCs was determined by X-gal staining. Culture expanded murine MSCs were CD31 and CD45 in vitro (data not shown), as reported by others.29 The supernatant was collected from genetically engineered cells, and mouse Epo secretion was assessed by photometric enzyme-linked immunosorbent assay (ELISA) specific for human Epo (Roche Diagnostics, Indianapolis, IN, USA). Formal institutional approval was obtained for animal studies performed in this project and all animals were handled under the guidelines promulgated by the Canadian Council on Animal Care.

Southern blot analysis

Genomic DNA was isolated from Epo-IRES-EGFP stably transduced primary murine MSCs, as well as from unmodified marrow stroma, utilizing the QIAamp DNA mini kit (Qiagen, Mississauga, ON, Canada). For the Southern blot analysis, 10 μg of genomic DNA was digested with EcoRV, separated by electrophoresis in 1% agarose, and transferred to a Hybond-N nylon membrane (Amersham, Oakville, ON, Canada). The probe was prepared by 32P radiolabeling of the EGFP complete cDNA utilizing a Random Primed DNA Labeling Kit (Roche Diagnostics) and was hybridized with the membrane. The blot was subsequently washed, and exposed to a Kodak X-Omat film.

Stroma implantation and blood sample analysis

For the intraperitoneal implantations of ‘free’ cells, Epo-IRES-EGFP-modified stromal cells were trypsinized, concentrated by centrifugation, and the various concentrations of 105, 106, 5×106 and 107 cells in 1 ml of serum-free RPMI media (Wisent) were injected into the peritoneum of four cohorts of 3–4 syngeneic C57Bl/6 mice. Control mice (n=5) were implanted with 107 IRES-EGFP-modified MSCs. For the subcutaneous implantations of ‘free’ cells, 4×106 Epo-IRES-EGFP-modified MSCs were resuspended in 500 μl of RPMI media and injected in the subcutaneous space of each of five syngeneic mice. Control mice (n=4) were generated by subcutaneous administration of 4×106 IRES-EGFP MSCs. For the subcutaneous implantations of Matrigel-embedded MSCs, 4×106 Epo-IRES-EGFP-modified MSCs were resuspended in 50 μl of RPMI media, mixed with 500 μl Matrigel™ (Becton Dickinson) at 4°C and implanted by subcutaneous injection in the right flank of three syngeneic C57Bl/6 mice. Matrigel, at body temperature, rapidly acquires a semisolid form. Control mice (n=4) were implanted with 4×106 Matrigel-embedded IRES-EGFP MSCs. In addition, 4×106 LacZ-Epo-IRES-EGFP MSCs mixed in Matrigel were implanted in another three mice. Control mice (n=3) received 4×106 LacZ-IRES-EGFP MSCs in Matrigel. For the shorter 4-week study, LacZ-Epo-IRES-EGFP-modified MSCs were likewise injected embedded in Matrigel at the various cell doses of 4, 0.5, and 0.25×106 MSCs in each of four mice. Control mice (n=4) were equally generated by implantation of 0.5×106 Lac Z-IRES-EGFP MSCs enclosed in Matrigel. As a positive control, four mice were administered subcutaneously 1000 U of human recombinant Epo (Eprex™, Janssen-Ortho Inc., North York, ON, Canada) mixed in Matrigel. For the subcutaneous implantation of MSCs embedded in a ‘human-compatible’ bovine type I collagen-based matrix, 4–5×106 Epo-IRES-EGFP-modified stromal cells suspended in 150 μl DMEM with 10% FBS were placed on a 1 cm2 piece of porous Collagen Matrix (Collagen Matrix, Inc., NJ, USA) in the well of a 24-well plate. The matrix became soaked and 15 min later, 800 μl of complete media was added to the well and the MSC-embedded collagen incubated overnight at 37°C with 5% CO2. The following day, one MSC-embedded collagen implant was surgically introduced into the subcutaneous space behind the neck of each of the five syngeneic C57Bl/6 mice anesthetized by isoflurane inhalation. Control mice (n=5) were implanted with 4–5×106 IRES-EGFP-modified MSCs embedded in Collagen Matrix and five additional negative control mice received only the collagen. Blood samples were collected from the saphenous vein with heparinized microhematocrit tubes (Fisher Scientific, Pittsburgh, PA, USA) prior to implantation and every 1 or more weeks postimplantation. Mice were monitored for up to 10 months. Hct levels and plasma mEpo concentrations were ascertained from blood samples. Specifically, Hcts were quantitated by standard microhematocrit procedure, and mEpo concentrations in plasma preparations were assessed by ELISA for human Epo (Roche Diagnostics).

Matrigel implant removal and processing

At 4 weeks postimplantation, one group of mice implanted with LacZ gene modified MSCs (ie LacZ-Epo-IRES-EGFP MSCs and LacZ-IRES-EGFP MSCs) embedded in Matrigel were killed and their systemic circulation flushed through the left ventricle with 15 ml of 4°C phosphate-buffered solution (PBS) and then with 15 ml of 4°C 2% paraformaldehyde (PFA). Matrigel implants were recovered and immersed in 2% PFA at 4°C for 24 h and in X-gal solution (5 mM K3Fe(CN)6, 5 mM K4Fe(CN)6·3H2O, 0.01% sodium deoxycholate, 2 mM MgCl2, 1 mM EGTA, and 1 mg/ml X-gal in PBS with 0.02% NP40) for 16 h. Samples were then fixed with 10% formalin, embedded in paraffin and sections of 3–4 μm were prepared. For immunohistochemical staining, specimens were deparaffinized in toluene and rehydrated. Endogenous peroxidase was blocked using 3% hydrogen peroxide followed by incubation with 5% bovine serum albumin with 5% goat serum or 5% donkey serum in PBS for 30 min. Sections were placed at 37°C with primary antibodies (polyclonal goat anti-mouse CD31 at 1:100), followed by biotin-conjugated secondary antibodies (donkey anti-goat IgG from Santa Cruz at 1:100, or goat anti-rabbit at 1:200 from BD Pharmingen), washed, and treated with avidin-peroxidase (ABC Elite kit, Vector Laboratories) for 30 min. DAB substrate (Vector Laboratories) was used for reaction development. Sections were counterstained with hematoxylin and eosin, visualized with an Olympus BX60 microscope, and digital images retrieved on a computer equipped with Image Pro software (Media Cybernetics).


  1. 1

    Maione D et al. Prolonged expression and effective readministration of erythropoietin delivered with a fully deleted adenoviral vector. Hum Gene Ther 2000; 11: 859–868.

  2. 2

    Descamps V et al. Erythropoietin gene transfer and expression in adult normal mice: use of an adenovirus vector. Hum Gene Ther 1994; 5: 979–985.

  3. 3

    Tripathy SK, Black HB, Goldwasser E, Leiden JM . Immune responses to transgene-encoded proteins limit the stability of gene expression after injection of replication-defective adenovirus vectors. Nat Med 1996; 2: 545–550.

  4. 4

    Tripathy SK et al. Stable delivery of physiologic levels of recombinant erythropoietin to the systemic circulation by intramuscular injection of replication-defective adenovirus. Proc Natl Acad Sci USA 1994; 91: 11557–11561.

  5. 5

    Svensson EC et al. Long-term erythropoietin expression in rodents and non-human primates following intramuscular injection of a replication-defective adenoviral vector. Hum Gene Ther 1997; 8: 1797–1806.

  6. 6

    Kessler PD et al. Gene delivery to skeletal muscle results in sustained expression and systemic delivery of a therapeutic protein. Proc Natl Acad Sci USA 1996; 93: 14082–14087.

  7. 7

    Snyder RO et al. Efficient and stable adeno-associated virus-mediated transduction in the skeletal muscle of adult immunocompetent mice. Hum Gene Ther 1997; 8: 1891–1900.

  8. 8

    Zhou S, Murphy JE, Escobedo JA, Dwarki VJ . Adeno-associated virus-mediated delivery of erythropoietin leads to sustained elevation of hematocrit in nonhuman primates. Gene Ther 1998; 5: 665–670.

  9. 9

    Bohl D et al. Improvement of erythropoiesis in beta-thalassemic mice by continuous erythropoietin delivery from muscle. Blood 2000; 95: 2793–2798.

  10. 10

    Beall CJ et al. Transfer of the feline erythropoietin gene to cats using a recombinant adeno-associated virus vector. Gene Ther 2000; 7: 534–539.

  11. 11

    Benihoud K, Yeh P, Perricaudet M . Adenovirus vectors for gene delivery [review] [63 refs]. Curr Opin Biotechnol 1999; 10: 440–447.

  12. 12

    Monahan PE, Samulski RJ . AAV vectors: is clinical success on the horizon? [review] [48 refs]. Gene Ther 2000; 7: 24–30.

  13. 13

    Moskalenko M et al. Epitope mapping of human anti-adeno-associated virus type 2 neutralizing antibodies: implications for gene therapy and virus structure. J Virol 2000; 74: 1761–1766.

  14. 14

    Lochmuller H et al. Immunosuppression by FK506 markedly prolongs expression of adenovirus-delivered transgene in skeletal muscles of adult dystrophic [mdx] mice. Biochem Biophys Res Commun 1995; 213: 569–574.

  15. 15

    Miller HI . Gene therapy on trial. Science 2000; 287: 591–592.

  16. 16

    Wadman M . NIH under fire over gene-therapy trials. Nature 2000; 403: 237.

  17. 17

    Jenks S . Gene therapy death – ‘Everyone has to share in the guilt’. J Nat Cancer Inst 2000; 92: 98–100.

  18. 18

    Donsante A et al. Observed incidence of tumorigenesis in long-term rodent studies of rAAV vectors [letter; comment]. Gene Ther 2001; 8: 1343–1346.

  19. 19

    Rizzuto G et al. Efficient and regulated erythropoietin production by naked DNA injection and muscle electroporation. Proc Natl Acad Sci USA 1999; 96: 6417–6422.

  20. 20

    Payen E et al. Improvement of mouse beta-thalassemia by electrotransfer of erythropoietin cDNA. Exp Hematol 2001; 29: 295–300.

  21. 21

    Rizzuto G et al. Gene electrotransfer results in a high-level transduction of rat skeletal muscle and corrects anemia of renal failure. Hum Gene Ther 2000; 11: 1891–1900.

  22. 22

    Maruyama H et al. Continuous erythropoietin delivery by muscle-targeted gene transfer using in vivo electroporation. Hum Gene Ther 2000; 11: 429–437.

  23. 23

    Kreiss P, Bettan M, Crouzet J, Scherman D . Erythropoietin secretion and physiological effect in mouse after intramuscular plasmid DNA electrotransfer. J Gene Med 1999; 1: 245–250.

  24. 24

    Jiao S et al. Direct gene transfer into nonhuman primate myofibers in vivo. Hum Gene Ther 1992; 3: 21–33.

  25. 25

    Tripathy SK et al. Long-term expression of erythropoietin in the systemic circulation of mice after intramuscular injection of a plasmid DNA vector. Proc Natl Acad Sci USA 1996; 93: 10876–10880.

  26. 26

    Roth DA et al. Non-viral gene transfer of blood coagulation factor VIII in patients with severe hemophilia A. Blood 2000; 96: 590 (abstract).

  27. 27

    Ding L et al. Bone marrow stromal cells as a vehicle for gene transfer. Gene Ther 1999; 6: 1611–1616.

  28. 28

    Clark BR, Keating A . Biology of bone marrow stroma [review] [57 refs]. Ann NY Acad Sci 1995; 770: 70–78.

  29. 29

    Deans RJ, Moseley AB . Mesenchymal stem cells: biology and potential clinical uses [review] [92 refs]. Exp Hematol 2000; 28: 875–884.

  30. 30

    Prockop DJ . Marrow stromal cells as stem cells for nonhematopoietic tissues [review] [32 refs]. Science 1997; 276: 71–74.

  31. 31

    Gerson SL . Mesenchymal stem cells: no longer second class marrow citizens [news; comment]. Nat Med 1999; 5: 262–264.

  32. 32

    Makino S et al. Cardiomyocytes can be generated from marrow stromal cells in vitro. J Clin Invest 1999; 103: 697–705.

  33. 33

    Reyes M et al. Purification and ex vivo expansion of postnatal human marrow mesodermal progenitor cells. Blood 2001; 98: 2615–2625.

  34. 34

    Horwitz EM et al. Transplantability and therapeutic effects of bone marrow-derived mesenchymal cells in children with osteogenesis imperfecta. Nat Med 1999; 5: 309–313.

  35. 35

    Horwitz EM et al. Clinical responses to bone marrow transplantation in children with severe osteogenesis imperfecta. Blood 2001; 97: 1227–1231.

  36. 36

    Keating A et al. Donor origin of the in vitro haematopoietic microenvironment after marrow transplantation in man. Nature 1982; 298: 280–283.

  37. 37

    Jaalouk DE et al. Glucocorticoid-inducible retrovector for regulated transgene expression in genetically engineered bone marrow stromal cells. Hum Gene Ther 2000; 11: 1837–1849.

  38. 38

    Chiang GG et al. Bone marrow stromal cell-mediated gene therapy for hemophilia A: in vitro expression of human factor VIII with high biological activity requires the inclusion of the proteolytic site at amino acid 1648. Hum Gene Ther 1999; 10: 61–76.

  39. 39

    Hurwitz DR et al. Systemic delivery of human growth hormone or human factor IX in dogs by reintroduced genetically modified autologous bone marrow stromal cells. Hum Gene Ther 1997; 8: 137–156.

  40. 40

    Keating A, Guinn BA, Laraya P, Wang XH . Human marrow stromal cells electrotransfected with human FIX cDNA engraft in SCID mice and transcribe human Factor IX. Exp Hematol 1996; 24: 1056 (abstract).

  41. 41

    Eliopoulos N, Crosato M, and Galipeau J . High-level erythropoietin production from genetically engineered bone marrow stroma implanted in non-myeloablated, immunocompetent mice. Blood 2000; 96: 802 (absract).

  42. 42

    Nolta JA, Hanley MB, Kohn DB . Sustained human hematopoiesis in immunodeficient mice by cotransplantation of marrow stroma expressing human interleukin-3: analysis of gene transduction of long-lived progenitors. Blood 1994; 83: 3041–3051.

  43. 43

    Brouard N et al. Transplantation of stromal cells transduced with the human IL3 gene to stimulate hematopoiesis in human fetal bone grafts in non-obese, diabetic-severe combined immunodeficiency mice. Leukemia 1998; 12: 1128–1135.

  44. 44

    Cherington V et al. Retroviral vector-modified bone marrow stromal cells secrete biologically active factor IX in vitro and transiently deliver therapeutic levels of human factor IX to the plasma of dogs after reinfusion. Hum Gene Ther 1998; 9: 1397–1407.

  45. 45

    Bartholomew A et al. Baboon mesenchymal stem cells can be genetically modified to secrete human erythropoietin in vivo. Hum Gene Ther 2001; 12: 1527–1541.

  46. 46

    Markowitz D, Goff S, Bank A . A safe packaging line for gene transfer: separating viral genes on two different plasmids. J Virol 1988; 62: 1120–1124.

  47. 47

    Galipeau J et al. Vesicular stomatitis virus G pseudotyped retrovector mediates effective in vivo suicide gene delivery in experimental brain cancer. Cancer Res 1999; 59: 2384–2394.

  48. 48

    Morgenstern JP, Land H . A series of mammalian expression vectors and characterisation of their expression of a reporter gene in stably and transiently transfected cells. Nucleic Acids Res 1990; 18: 1068.

  49. 49

    Regulier E et al. Continuous delivery of human and mouse erythropoietin in mice by genetically engineered polymer encapsulated myoblasts. Gene Ther 1998; 5: 1014–1022.

  50. 50

    Rinaudo D, Toniatti C . Sensitive ELISA for mouse erythropoietin. Biotechniques 2000; 29: 218–220.

  51. 51

    Naffakh N et al. Long-term secretion of therapeutic proteins from genetically modified skeletal muscles. Hum Gene Ther 1996; 7: 11–21.

  52. 52

    Lejnieks DV, Ramesh N, Lau S, Osborne WR . Stomach implant for long-term erythropoietin expression in rats. Blood 1998; 92: 888–893.

  53. 53

    Lin Y et al. Use of blood outgrowth endothelial cells for gene therapy for hemophilia A. Blood 2002; 99: 457–462.

  54. 54

    Ramesh N et al. High-level human adenosine deaminase expression in dog skin fibroblasts is not sustained following transplantation. Hum Gene Ther 1993; 4: 3–7.

  55. 55

    Palmer TD, Rosman GJ, Osborne WR, Miller AD . Genetically modified skin fibroblasts persist long after transplantation but gradually inactivate introduced genes. Proc Natl Acad Sci USA 1991; 88: 1330–1334.

  56. 56

    Cristofalo VJ, Volker C, Francis MK, Tresini M . Age-dependent modifications of gene expression in human fibroblasts [review] [261 refs]. Crit Rev Eukaryotic Gene Expression 1998; 8: 43–80.

  57. 57

    Gussoni E et al. Normal dystrophin transcripts detected in Duchenne muscular dystrophy patients after myoblast transplantation. Nature 1992; 356: 435–438.

  58. 58

    Mendell JR et al. Myoblast transfer in the treatment of Duchenne's muscular dystrophy [see comments]. New Engl J Med 1995; 333: 832–838.

  59. 59

    Morgan JE . Cell and gene therapy in Duchenne muscular dystrophy [review]. [104 refs]. Hum Gene Ther 1994; 5: 165–173.

  60. 60

    Osborne WR et al. Gene therapy for long-term expression of erythropoietin in rats. Proc Natl Acad Sci USA 1995; 92: 8055–8058.

  61. 61

    Clowes MM et al. Long-term biological response of injured rat carotid artery seeded with smooth muscle cells expressing retrovirally introduced human genes. J Clin Invest 1994; 93: 644–651.

  62. 62

    Lejnieks DV et al. Granulocyte colony-stimulating factor expression from transduced vascular smooth muscle cells provides sustained neutrophil increases in rats. Hum Gene Ther 1996; 7: 1431–1436.

  63. 63

    Mitani K, Wakamiya M, Caskey CT . Long-term expression of retroviral-transduced adenosine deaminase in human primitive hematopoietic progenitors. Hum Gene Ther 1993; 4: 9–16.

  64. 64

    Chertkov JL et al. The hematopoietic stromal microenvironment promotes retrovirus-mediated gene transfer into hematopoietic stem cells. Stem Cells 1993; 11: 218–227.

  65. 65

    Quesenberry PJ et al. Engraftment of normal murine marrow into nonmyeloablated host mice. Blood Cells 1994; 20: 348–350.

  66. 66

    Culver K et al. Lymphocytes as cellular vehicles for gene therapy in mouse and man. Proc Natl Acad Sci USA 1991; 88: 3155–3159.

  67. 67

    Koc ON et al. Culture-expanded autologous human mesenchymal stem cells (MSCs) circulate in blood and retain proliferative capacity following iv infusion into breast cancer patients. Blood 1998; 92: 274 (abstract).

  68. 68

    Lazarus HM et al. Ex vivo expansion and subsequent infusion of human bone marrow-derived stromal progenitor cells (mesenchymal progenitor cells): implications for therapeutic use. Bone Marrow Transplant 1995; 16: 557–564.

  69. 69

    Keating A, Berkahn L, Filshie R . A phase I study of the transplantation of genetically marked autologous bone marrow stromal cells. Hum Gene Ther 1998; 9: 591–600.

  70. 70

    Chuah MK et al. Long-term persistence of human bone marrow stromal cells transduced with factor VIII-retroviral vectors and transient production of therapeutic levels of human factor VIII in nonmyeloablated immunodeficient mice. Hum Gene Ther 2000; 11: 729–738.

  71. 71

    Allay JA et al. LacZ and interleukin-3 expression in vivo after retroviral transduction of marrow-derived human osteogenic mesenchymal progenitors. Hum Gene Ther 1997; 8: 1417–1427.

  72. 72

    Pereira RF et al. Marrow stromal cells as a source of progenitor cells for nonhematopoietic tissues in transgenic mice with a phenotype of osteogenesis imperfecta. Proc Natl Acad Sci USA 1998; 95: 1142–1147.

  73. 73

    Koc ON, Lazarus HM . Mesenchymal stem cells: heading into the clinic [review]. Bone Marrow Transplant 2001; 27: 235–239.

  74. 74

    Naffakh N et al. Sustained delivery of erythropoietin in mice by genetically modified skin fibroblasts. Proc Natl Acad Sci USA 1995; 92: 3194–3198.

  75. 75

    Rosenzweig M et al. Induction of cytotoxic T lymphocyte and antibody responses to enhanced green fluorescent protein following transplantation of transduced CD34+ hematopoietic cells. Blood 2001; 97: 1951–1959.

  76. 76

    Stripecke R et al. Immune response to green fluorescent protein: implications for gene therapy. Gene Ther 1999; 6: 1305–1312.

  77. 77

    Stupack DG et al. Apoptosis of adherent cells by recruitment of caspase-8 to unligated integrins. J Cell Biol 2001; 155: 459–470.

  78. 78

    Jiang Y et al. Pluripotency of mesenchymal stem cells derived from adult marrow. Nature 2002; 418: 41–49.

  79. 79

    Ruschitzka FT et al. Nitric oxide prevents cardiovascular disease and determines survival in polyglobulic mice overexpressing erythropoietin. Proc Natl Acad Sci USA 2000; 97: 11609–11613.

  80. 80

    Eliopoulos N et al. Human cytidine deaminase as an ex vivo drug selectable marker in gene-modified primary bone marrow stromal cells. Gene Ther 2002; 9: 452–462.

Download references


We thank Franca Pulice for expert technical support. N Eliopoulos is a fellow of the Leukemia Research Fund of Canada and J Galipeau is a recipient of the Medical Research Council of Canada Clinician-Scientist Award. This work is supported by the Bayer-Canadian Blood Services/Hema-Quebec Partnership Fund and the Canadian Stem Cell Network.

Author information

Correspondence to J Galipeau.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Eliopoulos, N., Al-Khaldi, A., Crosato, M. et al. A neovascularized organoid derived from retrovirally engineered bone marrow stroma leads to prolonged in vivo systemic delivery of erythropoietin in nonmyeloablated, immunocompetent mice. Gene Ther 10, 478–489 (2003). https://doi.org/10.1038/sj.gt.3301919

Download citation


  • retroviral vector
  • gene transfer
  • cell therapy
  • erythropoietin
  • marrow stroma

Further reading