WO2013186264A1

WO2013186264A1 - Immortalized mesenchymal stem cells that can be killed through an inducible apoptosis system

Info

Publication number: WO2013186264A1
Application number: PCT/EP2013/062162
Authority: WO
Inventors: Paul-Emile BOURGINE
Original assignee: Universitätsspital Basel
Priority date: 2012-06-12
Filing date: 2013-06-12
Publication date: 2013-12-19

Abstract

The invention relates to mesenchymal stem cell comprising a first transgene encoding a mammalian telomerase, and a second transgene encoding a fusion protein constituted of a monomer of a protein able to induce apoptosis in said mesenchymal stem cell in dimeric form, and a protein able to bind to a small molecule pharmaceutical drug, wherein said fusion molecule is able to dimerize and trigger apoptosis in said mesenchymal stem cell when said cell is exposed to said small molecule pharmaceutical drug. The invention further relates to methods of generating such cell, and its use in providing devitalized tissue matrices for transplantation.

Description

IMMORTALIZED MESENCHYMAL STEM CELLS THAT CAN BE KILLED THROUGH AN INDUCIBLE APOPTOSIS SYSTEM

Description

The present invention relates to methods to provide immortalized stem cells having a chemically inducible caspase system, such immortalized stem cells and preparations thereof, and to methods to generate devitalized tissue grafts.

Human bone marrow-derived Mesenchymal Stem/Stromal Cells (hMSCs) are defined as a cellular fraction positive for CD73, CD90, CD105, and negative for hematopoietic markers, while being able to stably differentiate in vitro into osteoblasts, adipocytes and chondrocytes [Pittenger et al., Science 1999;284:143-147]. The multidifferentiation capacity of hMSCs makes them promising candidates for regenerative medicine. The chondrogenic and osteogenic potential of hMSCs is used for the repair of damaged articular cartilage but also for bone tissue engineering by recapitulating intramembranous or endochondral ossification processes.

Approaches to translate the potential of hMSCs to clinical applications suffer from both the heterogeneity of this population and the lack of supply of suitable cells. The low amount of hMSCs in bone marrow extracts (0.001 to 0.1 % of nucleated cells) typically necessitates an in vitro expansion phase prior to use. This expansion phase is limited by the replicative senescence phenomenon occurring under in vitro culture conditions, after 30 to 40 population doublings (PD). Also, the differentiation potential of hMSCs displays significant variations from donor to donor, leading to standardization problems. Overall, the low frequency of hMSCs combined with their short life-span and the high inter-donor variability within hMSCs preparations limits a coherent exploitation of their therapeutic potential.

Immortalized hMSCs were developed by inserting of a human telomerase catalytic subunit (hTERT) [Jun et al., Cell Physiol Biochem. 2004; 14(4-6):261-8]. The generation of hTERT- MSCs lines allowed a significant extent of hMSCs life-span (>300PD) while preserving some of the properties of primary hMSCs. The effect of hTERT-driven immortalization on the tumour homing capacity of hMSCs has not been addressed and no study has reported the maintenance of immunomodulatory properties of human bone marrow-derived hMSCs line. Furthermore, safety concerns impede widespread clinical use of telomerase-expression based immortalized MSC lines. A strong telomerase expression has been associated with many cancer types. Moreover, neoplastic transformations can occur following hTERT- transformation of hMSCs,

Straathof et al. have published an approach whereby human T cells are modified to express caspase 9 fused to a small-molecule binding protein that, upon binding to a dimer of its small molecule ligand. The system induces cell death in over 99% of cells [Straathof et al., Blood 2005; 105:4247-4254]. The system has been applied to mesenchymal stem cells (Almeida Ramos et al. 2010, Stem Cells 2010;28:1 107-1 1 15). Expressed transgenes that induce cell death after exposure of the cell to a small molecule pharmaceutical agent are referred to as iDS (inducible death systems) in the following.

The need for organ or tissue transplants exceeds the current supply, due to the lack of donors and stringent requirements of immuno-compatibility between donor and recipient. To solve such limitations, one approach in tissue engineering makes use of the generation of decellularized tissue. After the in vitro generation of a graft, the decellularization proceeds to devitalization and the removal of the cellular fraction from the graft. The proteins within this remaining extra-cellular matrix (ECM) are shown to be well conserved, minimizing any immune-reaction by the recipient. This method allows the production of cell-free but functional tissues through the remaining ECM that retains key structural and/or instructive properties, such as osteo-inductive or angiogenic proteins. Acellular grafts can be theoretically implanted without immuno-compatibility concerns. This technique led to many applications in tissue engineering, such as the design of cell-free myocardial, vascular, skin, cartilage or bone graft.

The devitalization method is a key step in the generation of cell-free grafts. It aims at removing all cellular material without adversely affecting the composition, mechanical integrity but also the biologic activity of the remaining ECM that carries specific properties. Currently, physical treatment (freeze & thaw cycles, sonication, pressure, mechanical agitation), enzymatic (Trypsin) or chemical treatments (Sodium deoxycholate, Triton X solutions) are used to eliminate the living fraction from the generated graft. Those methods must maintain a tight balance of having an optimal preservation of the ECM properties while obtaining an efficient removal of the cellular component. Until now, no method was shown to both successfully remove the DNA from the graft and preserve the ECM.

Endochondral ossification has been proposed as a method to engineer osteogenic grafts. Those constructs are generated by the condensation and differentiation of hMSCs into chondrocytes. After 5 weeks of in vitro culture, a mature cartilaginous template surrounded by a calcified collar is obtained (Kawada et al. (2004). Blood 104:3581-3587). Following ectopic implantation in nude mice, the remodeling and vascularization of the engineered grafts leads to the formation of mature bone. Nevertheless, the implanted material contains a living fraction of hMSCs that can cause rejection of the graft in immuno-competent animals.

The objective of the present invention is to improve on the above state of the art to provide safe and efficacious means for the generation and application of hMSC in a variety of clinical approaches. This objective is attained by the subject matter of the independent claims. The present invention relates to human bone marrow-derived Mesenchymal Stem/Stromal Cells (hMSCs) that are immortalized by introduction of a telomerase transgene expression construct, and an inducible genetic construct leading to apoptosis upon induction.

hTERT-immortalized hMSCs including an iDS (inducible death system) provide an unlimited, safe and well-characterized cell source, thus solving limitations raised by the use of primary hMSCs, facilitating standardized clinical use of hMSCs. The examples provided herein demonstrate that both gene expression constructs function in the same cells and do not disrupt the normal differentiation and immune-modulatory capabilities of the hMSCs. The cells also do not appear to be tumorigenic when transplanted into immune deficient mice.

The engineered cell line of the examples was shown to conserve the properties of primary hMSCs, while being efficiently inducible toward apoptosis in vitro and in vivo. Combining immortalization and suicide device provides a safe and standardized hMSCs source.

According to a first aspect of the invention, a method to generate a mesenchymal stem cell preparation, particularly a human bone marrow-derived mesenchymal stem/stromal cell (hMSC) preparation, is provided. This method comprises the steps of

a. providing a cell preparation that comprises mesenchymal stem cells, particularly human bone marrow-derived mesenchymal stem/stromal cells (hMSCs), b. introducing, in an immortalization step, into said mesenchymal stem cells a first nucleic acid sequence encoding a mammalian telomerase under control of a first promoter sequence operable in said mesenchymal stem cells, yielding immortalized mesenchymal stem cells, and

c. introducing, in an iDS transduction step, into said immortalized mesenchymal stem cells a second nucleic acid sequence encoding a fusion protein comprising i. a monomer of an apoptosis-inducing protein, wherein said apoptosis- inducing protein is able to induce apoptosis in said mesenchymal stem cell when said apoptosis-inducing protein is present in dimeric form, and ii. a protein able to bind to a dimerizing small molecule pharmaceutical drug, iii. wherein said fusion molecule dimerizes and triggers apoptosis in said mesenchymal stem cell when said cell is exposed to said dimerizing small molecule pharmaceutical drug, and wherein

said second nucleic acid sequence is under control of a second promoter sequence operable in said mesenchymal stem cells.

In some embodiments, the starting material consists of primary human bone-marrow derived mesenchymal stromal cells. An example of a dimerizing small molecule pharmaceutical drug able to interact with the iDS is the "Chemical Inducer of Dimerization" (CID) shown in Keenan et al., Bioorganic &

Medicinal Chemistry 1998, 6, 1309-1335. Other examples are discussed in Kley, Chemistry & Biology 2004, 1 1 , 599-608. One frequently used CID molecule is AP1903 (W09731898A1 ; CAS No. 195514-63-7; 2-piperidinecarboxylic acid, 1 -[(2S)-1-oxo-2-(3,4,5-trimethoxyphenyl) butyl]-, 1 ,2-ethanediylbis[imino(2-oxo- 2, 1 -ethanediyl)oxy-3, 1 -phenylene[( 1 R)-3-(3,4- dimethoxyphenyl)propylidene]] ester (2S,2'S) ).

An example of a molecule able to bind to a dimerizing small molecule pharmaceutical drug is the FK506-binding protein family of polypeptides. One example is FKBP12, GenBank AH002 818. In some embodiments, a short peptidic linker is inserted between the amino acid sequence of the apoptosis-inducing protein and the protein that is able to bind the dimerizing drug. An example of this system is shown in Straathof et al., ibid..

In some embodiments, said first and/or second nucleic acid sequence is introduced by retroviral transduction. In some embodiments, said first nucleic acid sequence is introduced into said mesenchymal stem cells by lentiviral transduction. Both lentiviral and retroviral methods allow reaching high transduction efficiency in mesenchymal stem cells.

Lentiviral/retroviral gene transfer also allows for a constitutive expression of the transgene, as inserted into the genome of the target cells which is transmitted to the daughter cells.

In some embodiments, subsequent to the immortalization step or subsequent to the iDS transduction step, a single clone is selected for further propagation.

In some embodiments, said mammalian telomerase is a human telomerase. In some embodiments, said mammalian telomerase is the telomerase reverse transcriptase as defined in GenBank BAC1 1010 (Gene ID 7015).

In some embodiments, said first and/or said second promoter sequence is constitutively expressed. In some embodiments, said first and/or said second promoter sequence is a cytomegalovirus (CMV) promoter or SV40 promoter, particularly the CMV immediate early promoter or SV40 early promoter.

In some embodiments, said protein able to induce apoptosis is a caspase, particularly caspase 9 (Uniprot ID P5521 1 ). In some embodiments, said protein able to induce apoptosis is a caspase of the apoptotic pathway selected from caspase-3, caspase-7, caspase-8, caspase-10, caspase-2, caspase-6. In an alternative embodiment, apoptosis is induced by triggering of death-receptors, as described below (kiss-of-death strategy).

In some embodiments, said first and/or said second nucleic acid sequence comprises an expressed indicator transgene (by way of example: a fluorescent gene (EGFP) or a surface marker (CD19)) facilitating the selection of cells expressing said expressed indicator transgene. The expressed indicator transgene facilitates quality control of the resultant cells, or enables selecting for populations consisting essentially only of cells expressing the iDS system.

In some embodiments, said human bone marrow-derived mesenchymal stem/stromal cell preparation comprises at least 85%, 90% or 95% of cells that are CD73, CD90 and / or CD105 positive.

A cells is positive for a certain marker, such as one of the mentioned CD (cluster of differentiation) markers, in the context of the present specification, if a cell shows a significantly higher signal, for example a fluorescence signal, if reacted with a fluorescence- based antibody, compared to a cell commonly known as "negative" for the marker.

In some embodiments, the cells of the invention are separated by means of magnetic separation. This is a process known to the skilled artisan, whereby cells are separated in a very strong magnetic field by retention of the antigen positive cells by antibody-coupled magnetic particles. The cells of the invention can be separated e.g. by retention mediated by magnetic particles coupled to antibodies specific for cell surface molecules expressed on the surface of these cells, such as CD73, CD90 and/or CD105.

Means and methods for separation of cells regarding their expression of antigens are known to the skilled artisan. According to one embodiment of the present invention the cell preparation is obtained using a method which, as an alternative or in addition to the separation with respect to CD90 positivity, comprises further separation steps, which select regarding further possible characteristics of the cell preparation:

According to an embodiment of the present invention the cell preparation can be

characterized in that a plurality of the cells contained therein is CD105 positive. This means that at least 50% of the cells contained in this population are stained in the FACS by means of a standard dye marker, for instance the dye marker stated in the examples. CD105 is a typical marker for endothelial cells and mesenchymal cells. Preferably, according to an embodiment of the present invention, the cell preparation is at least 90%, more preferably at least 98% CD105 positive.

In some embodiments, said human bone marrow-derived mesenchymal stem/stromal cell preparation comprises at least 85%, 90% or 95% of cells that are positive for CD73, CD90 and CD105.

In some embodiments, the starting cell preparation that comprises mesenchymal stem cells, particularly human bone marrow-derived mesenchymal stem/stromal cells (hMSCs) (step a. above) is a cell line. In some embodiments, subsequent to the immortalization step and/or the iDS transduction step, said cell preparation is submitted to a sorting step whereby cells expressing said indicator transgene are selected.

According to another aspect of the invention, a mesenchymal stem cell, particularly a human bone marrow-derived mesenchymal stem/stromal cell (hMSC), is provided. Said hMSC comprises

a. a first transgene comprising a first nucleic acid sequence encoding a mammalian

telomerase under control of a first promoter sequence operable in said mesenchymal stem cell, and

b. a second transgene comprising a second nucleic acid sequence encoding a fusion protein comprising

i. a monomer of an apoptosis-inducing protein, wherein said apoptosis-inducing protein is able to induce apoptosis in said mesenchymal stem cell when said apoptosis- inducing protein is present in dimeric form, and

ii. a protein able to bind to a dimerizing small molecule pharmaceutical drug, iii. wherein said fusion molecule dimerizes and trigger apoptosis in said mesenchymal stem cell when said cell is exposed to said dimerizing small molecule

pharmaceutical drug, and

said second nucleic acid sequence is under control of a second promoter sequence operable in said mesenchymal stem cell.

In some embodiments, the mesenchymal stem cell of the invention is characterized in that

- the cell is positive for CD73, CD90 and / or CD105;

- the cell is positive for CD44, CD29, and/or CD73;

the cell is negative for hematopoietic (CD34, CD45) and epithelial markers (Epcam, E- cadherin);

said mammalian telomerase is a human telomerase, particularly telomerase reverse transcriptase as defined in GenBank BAC1 1010;

- said protein able to induce apoptosis is a caspase, particularly caspase 9 (Uniprot ID P5521 1 ); and/or

- said first and/or said second nucleic acid sequence comprises an expressed transgene facilitating the selection of cells expressing said expressed transgene (by way of example: a fluorescent gene (EGFP) or a surface marker (CD19)).

According to yet another aspect of the invention, a cell preparation comprising more than 80%, 90%, 95%, or 98% of mesenchymal stem cells according to the above aspect of the invention is provided. Alternatively, a cell preparation generated by a method according to the invention as defined above is provided.

In some embodiments, more than 70%, 80%, 90%, 95%, or 98% of the cells of the preparation:

a. are positive for expression of CD73, CD90 and / or CD105; and or

b. undergo apoptosis upon exposure to said small molecule pharmaceutical drug; and/or c. are positive for expression of said expressed indicator transgene.

According to yet another aspect of the invention, a method for generating a tissue matrix is provided, comprising the steps of:

a. providing a support matrix, and

b. contacting said support matrix with a cell preparation derived by a method of the

invention as laid out above, or a cell preparation according to the invention, in an ex- vivo cultivation step.

In some embodiments, the method for generating a tissue matrix further comprises the step of exposing said matrix to a dimerizing small molecule pharmaceutical drug capable of inducing apoptosis in said cell preparation. Examples for such molecules are discussed above.

In some embodiments, the ex-vivo cultivation step is performed under conditions suitable for osteogenic, adipogenic or chondrogenic differentiation.

In one embodiment, the ex-vivo cultivation step is performed under conditions suitable for adipogenic differentiation, keeping cells in DMEM complete medium during a first cycle, and keeping cells in alternating cycles of exposure to a first adipogenic differentiation medium comprising dexamethasone (CAS No. 50-02-2), indomethacin (CAS No. 53-86-1 ), insulin and 3-isobutyl-1 -methylxanthine (IBMX, CAS No. 28822-58-4) during three days, and subsequent exposure to a second adipogenic differentiation medium comprising insulin but no

dexamethasone, indomethacin or IBMX during one day, said cycles being repeated between 2 and 6 times, preferably 4 times.

In one embodiment, the ex-vivo cultivation step is performed under conditions suitable for osteogenic differentiation, keeping cells in alpha-MEM complete medium supplemented with dexamethasone, L-ascorbic acid-2-phosphate and β-glycerophosphate.

In one embodiment, the ex-vivo cultivation step is performed under conditions suitable for chondrogenic differentiation, keeping cells in DMEM complete medium supplemented with dexamethasone, L-ascorbic acid-2-phosphate andTGFbl (transforming growth factor beta 1 (Uniprot ID P01 137).

In some embodiments, the method is performed in a 3D perfusion bioreactor.

Perfusion bioreactors are bioreactors allowing the perfusion of culture medium through the pores of the used scaffold, using an oscillating system. Cell seeding of scaffolds (i.e., the loading of isolated cells into the pores of a scaffold) is the first step in establishing a 3D culture and can play a crucial role in determining the progression of 3D tissue formation. However, conventional methods of cell seeding typically yield a non-uniform distribution of cells within a scaffold, which may subsequently lead to inhomogeneous tissue development. Subsequently, when cell-seeded constructs are cultured using conventional methods (i.e., statically cultured in Petri dishes, the resulting engineered tissues typically contain necrotic cells within the internal region of the tissue graft, surrounded by a dense layer of cells only at the construct periphery. As an alternative, it has been demonstrated that by perfusing a cell suspension directly through the pores of a 3D scaffold in a perfusion bioreactor system, cells can be reproducibly and uniformly distributed throughout the scaffold volume (Wendt et al.- Biotechnology and bioengineering 2003, Davisson et al.-J Orthop Res. 2002), establishing a template for uniform tissue development. Prolonged perfusion culture of the cell-seeded constructs in a perfusion bioreactor subsequently supports the development of viable and uniform tissues.According to yet another aspect of the invention, a tissue matrix obtainable by a method for generating a tissue matrix of the invention is provided.

Similarly, a devitalized tissue matrix obtainable or obtained by such a method is provided.

In some embodiments, the tissue matrix is a devitalized tissue matrix consisting essentially of a mature hypertrophic cartilage template.

In some embodiments, the tissue matrix is a devitalized tissue matrix consisting essentially of a bone tissue template.

In some embodiments, the tissue matrix is a devitalized tissue matrix consisting essentially of a skin template.

The present invention provides a novel devitalization method by the use of an inducible death system. Following the iDS device deployment, for example by integration using a retro- or lentivirus, the activation of this system in the transduced target cells leads to a high killing efficiency (>90%) while having the advantage of leaving the produced ECM intact. To investigate the general benefit of this method, the examples of the present specification compare the freeze & thaw method to the iDS method in the context of devitalized endochondral bone graft generation. A strategy adopted by the present inventors was to devitalize endochondral constructs prior implantation in order to eliminate the immunogenic cells, preserving the secreted ECM. The preservation of this matrix is crucial as it contains proteins (MMP13, BMP-2) important for the recruitment of host cells that participate to the remodeling and bone formation. Other proteins, such as the VEGF family, are preserved in some embodiments, facilitating the vascularization and successful engraftment of the constructs. The prior art methods employ the freeze & thaw devitalization method, which compromises the structure and function of the ECM. The present specification provides the proof of principle that an iDS devitalization method reaches a similar or superior killing of the cells while better preserving the ECM, thus leading to a more efficient bone formation.

The method used to implement the apoptotic concept for ECM decellularization will critically determine the killing efficiency and overall success of the strategy. Different approaches are feasible to achieve this aim, which notably differ according to the type of pathway (i.e., intrinsic vs. extrinsic) to be activated. Alternative or supplementary ways to induce apoptosis, useful for the above methods of the invention:

Kiss-of-death

This strategy entirely relies on the extrinsic pathway activation through the delivery of specific ligands that bind their corresponding death receptors (DRs) of the TNF superfamily (Table 1 ).

Table 1 *: all ID numbers in brackets refer to Uni-Prot ID Nos.

As the DRs are ubiquitously expressed, the activation of the extrinsic apoptotic pathway can be considered for the decellularization of virtually any tissue/organ. Nevertheless, the choice of the ligand, and consequently the type of DR to target, has to be carefully considered. While some ligands are known to specifically activate the programmed cell death, others (DR-1 , DR-3, DR-6 and EDAR) may have also anti-apoptotic effects, leading to the activation of survival signals. Furthermore, each cell type might have a different sensitivity to each ligand, since a differential expression of the DRs is observed from tissue to tissue. Some cells may also be more prone than others to escape apoptosis through the survival pathway. For instance, TRAIL was shown to induce apoptosis in cancer cells, but not in normal cells. Therefore, when elaborating a decellularization strategy, the type of DR to target should be selected according to the specific cellular system.

A certain number of pro-apoptotic ligands were already reported to successfully induce apoptosis in a variety of cell types. Human bone marrow-derived mesenchymal stromal cells (hMSC) were shown to be sensitive to FasL, in both an undifferentiated and differentiated status. FasL was also shown to induce apoptosis of cardiomyocytes and epidermal cells. Moreover, it is abundantly expressed in a variety of organs, such as the heart, kidney, pancreas and the liver. The delivery of TNFa was described as an efficient inducer in lung and intestinal epithelial cells. This factor also promotes chondrocyte and renal endothelial cell apoptosis.

This non-exhaustive list demonstrates the diversity of approaches employing DR ligands as apoptotic inducers. However, none of the published methods aimed at optimizing the observed killing. Optimization of this process requires screening for the appropriate ligand, for a suitable concentration and time of induction. Such optimization is within the knowledge of the skilled person. One possible optimization strategy relies on the combination of ligands, to generate a "customized death cocktail". This can be used to obtain synergistic effects; as for example one ligand primes cells to increase their sensitivity to a second one. As an example, it was shown that treating lung fibroblast and hepatocytes with TNFa sensitizes them to FasL induced apoptosis.

A potential limit of the Kiss-of-death strategy possibly involves the inflammatory effect of some ligands. A certain quantity of those factors may remain entrapped in the ECM, requiring additional rinsing for their removal from the graft. Furthermore, depending on the required concentrations of the selected ligands, the process may become costly.

Lethal-environmental-conditioning

Upon environmental stress, cells can naturally undergo apoptosis by activation of the intrinsic "mitochondrial" pathway. An apoptotic response could thus be induced by modulating environmental factors, including temperature, pH, as well as C0₂/0₂, nitric oxide (NO), and H₂0₂ content. As opposed to the physical freeze & thaw method, leading to a necrotic response, induction of apoptosis by temperature changes requires low variations, in either hyperthermic or hypothermic ranges. Ideally, temperature fluctuations from 10°C to 45°C are advised, whereby the effect on the matrix will also depend on the incubation time. NO has been described as a potent inducer of mitochondrial apoptosis, especially for

cardiomyocytes, pancreatic cells and chondrocytes. The use of hypoxic conditions was also reported to activate the apoptotic program in pancreatic cells or cardiomyocytes.

The advantages of the lethal-environmental-conditioning strategy rely on the minimal cost and simplicity of the set up required to modulate and control the operating conditions. Death-engineering

This strategy relies on the activation of any of the two apoptotic pathways by the use of a genetic approach. Since it requires the genetic engineering of cells prior to the generation of a tissue, this strategy relates mainly to the decellularization of engineered grafts. Apoptosis activation could in principle be achieved by modulating the expression level of key genes involved in the pathway. Nevertheless, since apoptosis implies phosphorylation and dimerization of specific molecular players, it cannot be triggered by simply overexpressing or silencing key genes of the transduction pathway.

One possibility to engineer cells to death consists in over-expressing death-receptors at the cell surface. This was shown to enhance cell sensitivity to the death-inducing ligand and could thus be used in conjunction with the Kiss-of-death strategy.

An alternative option relies on the implementation of a toxic transgene (e.g. apoptin, lectin), whose expression is under strict control. In this regard, the use of a tight and inducible expression system is a requirement in order to activate cell-suicide post-tissue generation and to avoid an excessive "leakiness" that may also result in premature cell-death. The stable integration of such a genetic construct is also critical, as existing transient expression systems were shown to persist only from days to a few weeks within the cells. Considering modified cells would be induced towards death after the synthesis of a tissue, which implicates an extensive culture time, transient systems may result in a poor killing due to the non-persistence of the genetic construct.

As an example for this approach, a genetic device addressing the above mentioned requirements is described in the example section contained herein. The system shows a high killing efficiency in transduced cells. This device is based on the constitutive expression of a modified caspase 9, whose dimerization can be activated through the delivery of a clinically approved inducer. This genetic approach was originally developed to improve the safety of cell-based therapy, but can also be used to induce the decellularization of a tissue. The efficiency of killing was already demonstrated in primary and differentiated hMSC. The advantage of this system relies in the absence of leakiness and the downstream activation of the apoptotic pathway, thus avoiding activation of the survival pathway.

The three aforementioned strategies (Kiss-of-death, lethal environmental conditioning and death-engineering) differ in protocol but are fully compatible and their combination may in fact synergize the final cell killing efficiency.

Tissue/Organ decellularization within perfusion systems

Devitalization can be considered as the first step of a decellularization procedure, ultimately aiming not only at killing, but also at removing the cellular fraction. Typical methods to eliminate the cell debris are based on extensive rinsing but risk to either lead to mechanical disruption of the tissue or to a non-efficient removal of the immunogenic material. Following the elegant concepts of whole organ perfusion developed in the recent years, here the inventors propose the use of perfusion bioreactor systems combined with induction of apoptosis in order to achieve an efficient and controlled tissue decellularization.

The use of a bioreactor system allows for a superior control of the process parameters, such as temperature or gas content, which is directly relevant for the lethal-environmental- conditioning strategy. Moreover, the perfusion enhances the killing efficiency by increasing the convection of the pro-apoptotic factors throughout the graft, especially while considering the Kiss-of-death or Death engineering strategies. The use of perfusion systems also plays a role in the efficient and controlled removal of the previously killed cellular component.

Controlling the flow patterns and associated induced shear is of advantage during the washing step in order to eliminate the apoptotic bodies from the ECM as they form, thus allowing for a non-invasive, yet efficient and standardized wash-out of any cellular material. The use of perfusion-based bioreactor systems, forcing a cell suspension or culture medium directly through the pores of a scaffold material, was previously shown to support the engineering of tissues with superior properties than typical static cultures. Thus, in some embodiments of the invention, a streamlined manufacturing process for off-the-shelf decellularized grafts is provided, whereby the same perfusion system is used first to develop the tissue and subsequently to decellularize the deposited ECM. Additional features and benefits of such paradigm, derived from other biotechnology settings, include automation, standardization, control and cost-effectiveness of the process implementation.

Tissues engineered to death: a proof of principle

In the context of bone regeneration, one attractive strategy is to generate decellularized grafts with osteo-inductive properties by decoration of materials with a cell-laid ECM. In contrast to the delivery of over-doses of single morphogens (e.g., defined bone

morphogenetic proteins), associated with cost and safety concerns, this approach relies on the embedding and presentation by the ECM of a cocktail of different factors in more physiological concentrations to activate a regenerative process (e.g., by acting on resident osteoblastic and endothelial lineage cells). In this context, human bone marrow-derived Mesenchymal Stromal Cells (hMSC) have already been used in the attempt to generate ECM with osteo- inductive properties, as a niche for reseeded MSC or for priming of endogenous progenitor cells. So far, however, the ECM deposited during in vitro culture could not be shown to be capable of inducing ectopic in vivo bone formation if deprived of the living cellular component. Combining the approaches of decellularization by apoptosis with a 3D perfusion culture system (Wendt et al., Biotechnol Bioeng 2003 Oct 20;84(2):205-14), together with the utilization of a suitable cell source, offers the opportunity to enhance the osteo-inductive properties of decellularized ECM. The present inventors successfully generated an immortalized hMSC line with properties similar to the original primary cells and controlled survival. The cell line maintains the capacity to differentiate towards the

osteogenic lineage, is not tumorigenic and - thanks to the implementation of a genetic device - can be pushed to programmed cell death by the delivery of a clinically approved chemical inducer.

Results shown herein demonstrate that the cell line could deposit a mineralized ECM in 2D culture and still be efficiently induced towards apoptosis (>95%, Figure 34). Moreover, when seeded on a porous ceramic scaffold within a 3D perfusion-based bioreactor system, the engineered cell line was capable to adhere, proliferate and deposit an ECM. The generated constructs could be directly and efficiently decellularized by the direct perfusion of the apoptotic-inducer through the ECM, leading to the successful generation of ECM-decorated, cell-free materials. As compared to the living counterparts, the "apoptized" tissues

maintained the amount of total collagen, as representative ECM protein, in contrast to those decellularized by a conventional freeze/thaw protocol, where a marked loss of total collagen content (74%) was measured (Figure 34). This setting thus provides a proof-of-principle for the decellularization of ECM produced by osteogenic cells following the described apoptotic Death-engineering strategy, in a perfusion bioreactor setup.

Further studies are required to assess (i) to which extent the apoptotic treatment may better preserve the deposited ECM, beyond total collagen, as compared to traditional protocols of decellularization and (ii) whether the resulting "apoptized" grafts have osteo-inductive properties. The paradigm could obviously be extended to other cell lines, for the deposition of a customized ECM specifically competent to induce regeneration of different tissues.

The invention is further characterized by the following examples and figures, from which further features, advantages or embodiments can be derived. The examples do not limit but illustrate the invention.

Description of the figures

Figure 1 : Functional map of lenti-hTERT-GFP (Biogenova, LG508). The system comprises a human telomerase gene and a GFP reporter gene, respectively under the control of the CMV and EF1 a promoter.

Figure 2: Transduction efficiency of the lenti-hTERT-GFP on primary hMSCs. After one round of transduction (left), more than 78% of the population expressed the transgene. The positive cell fraction was then sorted to reach a high degree of purity (>94%). X-axis: fluorescence intensity (log scale); y-axis: count. Figure 3: Telomerase activity of primary hMSCs and the immortalized population (Trap assay). The immortalized cells (196hT) displayed a higher telomerase activity (4 to 5 fold) than primary 196. Despite extensive doublings (>100PD), the telomerase activity of the immortalized population remains stable.

Figure 4: Population doubling levels of primary (red) and immortalized hMSCs (black). After 35 doublings, the primary cells stopped to proliferate. The immortalized population underwent more than 270 PD after one year of culture so far.

Figure 5: β-galactosidase assay of primary 196 (35PD) and the immortalized 196hT (40PD) population. Primary 196 were shown to degrade β-galactosidase (blue/dark coloration) when reaching 35PD, indicating an entry in a replicative senescence phase. The immortalized population bypassed the senescence-associated crisis and continues to proliferate.

Figure 6: Functional map of the iCasp9-ACD19 retrovector. The device consists in a modified FKBP12 binding domain linked to a caspase 9. The 2A-like sequence ensures the cleavage of the CD19 surface marker from the inducible caspase 9, after transduction of the mRNA. The complete device construction is detailed in Straathof et al., Blood 2005 (ibid.).

Figure 7: CD19 expression of the 196hT-iDS cell line. A pure population (>98%) could be sorted after CD 19 labeling of the successfully transduced cells, resulting in the isolation of the 196hT-iDS population. Axes as in Fig. 2.

Figure 8: Functional assessment of the inducible death system implemented in the 196hT- iDS population. After activation of the system by an overnight exposure to CID, more than 99% of the cells were killed (positivity for Annexin-V and/or propidium iodide (PI)). X-axis: fluorescence intensity PI (log scale); y-axis: fluorescence intensity of annexin-V staining (log scale).

Figure 9: Alizarin red staining of the 24 clones after 3 weeks of culture in osteogenic medium. The 196hT-iDS cell line was cultured in complete medium (CM) and served as negative control. Among all the clones, the clone 20 was shown to be the most potent in depositing a mineralized matrix. All pictures were taken with a similar objective and showed a representative area of the well.

Figure 10: Phenotypic analysis of the primary cells and the generated cell lines. The generated cell lines stably expressed typical markers of hMSCs (CD44, CD29, CD73, CD90) while being negative for hematopoietic (CD34, CD45) and epithelial markers (Epcam, E- cadherin). The GFP and CD 19 expression respectively reflect the successful immortalization and iDS insertion. Figure 11 : Alizarin red staining of M-SOD cells after 3 weeks of culture in complete (CM) or osteogenic medium (OM). Only in presence of osteogenic medium, cells were able to depositmineralized matrix (red nodules).

Figure 12: Total calcium deposition of M-SOD line during osteogenic differentiation. Cells were cultured for 3 weeks in either osteogenic medium (OM) or complete medium (CM). A strong calcium deposition is measured in osteogenic culture condition (>50pg/ml_) whereas almost no calcium was deposited during culture in complete medium (<0.5pg/ml_).

Figure 13: Gene expression levels of key osteoblastic genes after 14 days of culture in osteogenic medium. As compare to non-differentiated cells (expanded cells), a significant upregulation of Alkalyne Phosphatase (ALP), Bone Morphogenetic Protein 2 (BMP-2), Osteocalcin (OC) and Bone Sialoprotein (BSP) was observed.

Figure 14: Immunocytometry of Alkalyne Phosphatase and Osteocalcin proteins of M-SOD after 2 and 3 weeks of osteogenic differentiation (flow cytometry analysis). A temporal protein expression of ALP (>45%, at 2 weeks) and OC (>20% at 3 weeks) was observed in osteogenic conditions, whereas in complete medium condition few cells (<5%) were shown to expressed those proteins.

Figure 15: Oil red-0 staining of M-SOD after 3 weeks of culture in complete (CM) or adipogenic medium (AM). In presence of adipogenic medium, the M-SOD cell line was successfully differentiated into adipocytes as demonstrated by the positive staining of lipid droplets.

Figure 16: PPARy gene expression level after 3 weeks of culture in complete or adipogenic medium. In adipogenic conditions, a significant overexpression of PPARy mRNA level was detected (15.8 fold), thus confirming the capacity of M-SOD to differentiate into adipocytes.

Figure 17: Alcian blue staining of the primary 196, M-SOD cells and primary 199 after 2 weeks of culture. Both primary 196 and M-SOD cells failed in generating a cartilaginous matrix (blue staining). Primary 199 (hMSCs donor 199) served as positive control.

Figure 18: In vivo bone formation of the M-SOD cell line. Cells successfully formed bone nodules (B) in contact with the ceramic granules (Ce), as demonstrated by Masson's trichrome staining (up). Osteoblasts (Ob) can be identified within those nodules, within lacuna. Analysis by fluorescence microscope (down) revealed the expression of GFP within the cells surrounding the nodules, thus strongly suggesting a human origin of the bone

Figure 19: Immunomodulation assay (CFSE) of the M-SOD line. On top (A), percentage of proliferating CD4+ after 4 days of culture, with different ratio of M-SOD cells (MSCs), +/- PHA activation. Down (B), peaks of fluorescence in the activated CD4+ and activated CD4+ 1 :5 MSCs conditions, 4 days post PHA activation. Figure 20: Induction of death during osteogenic differentiation of the M-SOD cell line. Cells were induced toward death (+CID) after 1 , 2, or 3 weeks of culture in osteogenic medium (OM). Following exposure to CID, more than 95% of the cells became Annexin-V/PI positive, independently of their differentiation state.

Figure 21 : Induction of death during adipogenic differentiation of the M-SOD cell line. Cells were efficiently induced toward death (+CID) after 1 , 2 or 3 weeks of differentiation (>95%) in adipogenic medium (AM). A high killing efficiency (>95%) was reached despite the differentiation of the cells.

Figure 22: Gene expression levels of key oncogenes in expanded primary 196 and M-SOD. The primary hMSCs (primary 196) and M-SOD expressed very similar levels of p53, p21 , Rb1 and c-Myc.

Figure 23: Tumorigenicity assay of the M-SOD cell line. One millions of M-SOD cells expressing luciferase were injected into the right flank of NOD/SCI D mice (triplicate). Two days post cell injection, the luciferase system allow the localization of the M-SOD cells. After 8 days, the strength of the signal significantly decreased and was undetectable at day 16. More than 50 days post-injection, M-SOD cells remain undetectable so far demonstrating their non-tumorigenicity.

Figure 24: Experimental plan for the comparison of freeze & thaw (F&T) and apoptosis devitalization (+CID).

Figure 25: Generation of the 202-iDS population. After transduction and sorting of the CD19- positive fraction, more than 96% of the isolated cells carry the iDS.

Figure 26: Devitalization of the hypertrophic constructs. While a majority of cells remain alive (> 83%) in the non-devitalized constructs, both F&T and CID induction led to a high killing efficiency (superior to 90% and 93% respectively).

Figure 27: Safranin-0 staining of the generated hypertrophic construct after 5 weeks of in vitro culture. The 202 and 202-iDS cells successfully generated a good cartilage template, as assessed by the red staining of glycosaminoglycan (GAG). The two devitalization methods seem to not affect the matrix GAG content.

Figure 28: Alizarin red staining of the generated hypertrophic construct after 5 weeks of in vitro culture. All constructs displayed an outer mineralized rim typical of hypertrophic constructs which seems to not have been damaged by the devitalization techniques.

Figure 29: Gene expression levels of ECM, chondrogenic and hypertrophic related markers. All conditions show a significant up-regulation of those genes independently of the transduction (202 vs 202-iDS) or the devitalization step (F&T vs +CID), as compare to the level of expanded cells (dash line). No expression was detected in expanded cells regarding the Collagen II and Collagen X markers

Figure 30: Masson's trichrome staining of the samples retrieved 12 weeks post in vivo implantation. The non-devitalized samples (202 and 202-iDS) were remodeled into mature bone. The devitalization by F&T did not lead to bone formation while the +CID samples display a cortical bone formation but also the presence of bone marrow. B (bone), M (marrow), pB (perichondral bone).

Figure 31 : Macroscopic view of samples vascularization at explantation. The F&T construct did bot displayed signs of vascularization while the +CID samples showed clear evidence of vessel connection (presence of blood tissue).

Figure 32: Immunofluorescence analysis of devitalized constructs. Samples were stained for DAPI (nucleus) and CD31 (endothelial marker). Almost no blood vessels (CD31 positive) were detected in the F&T construct whereas the devitalization by apoptosis led to a better vascularization.

Tissue decellularization by apoptosis induction within 3D perfusion bioreactor. Three different approaches are proposed in order to decellularize a tissue by apoptosis-induction; Kiss-of-death (1 ), Lethal Environmental Conditioning (2), Death-engineering (3). The decellularization procedure within a 3D perfusion bioreactor system increases the convection of the apoptotic inducer, establishes controlled environmental conditions and facilitates the washout of apoptotic bodies. In particular, when considering the decellularization of engineered tissues, this system could be the basis for a streamlined process to generate an ECM and subsequently induce apoptosis within a single, closed device.

Generation of ECM-decorated materials by apoptosis-driven decellularization. A hMSC cell line was generated and transduced with an inducible caspase 9 gene (iCaspase 9). Cells were shown to deposit a mineralized ECM (Alizarin-red staining) in Petri dishes (2D) while being efficiently inducible toward apoptosis (>95% positivity for Annexin-V and/or PI). After perfusion-mediated seeding on porous ceramic scaffolds (3D), the cell line was capable to adhere, colonize the scaffold (MTT staining) and deposit a collagen-rich ECM. The apoptotic decellularization performed within the bioreactor system led to the generation of an ECM- decorated material without a measurable reduction of the total collagen content. As a comparison, the freeze&thaw method applied to the same constructs resulted in a collagen loss of 74% (final content: 5.61 ±1.06 pg/scaffold). These findings provide a proof-of-principle for the decellularization of ECM produced by osteogenic cells following the described apoptotic Death-engineering strategy, in a perfusion bioreactor setup. Alternative pathways towards the design of customized, decellularized ECM

(A) Cells can be engineered (1A) to secrete specific factors/morphogens. Following 3D culture, possibly in a bioreactor system and subsequent apoptosis-driven decellularization, the produced tissues are enriched in specific morphogens (2A). The differences in

composition are expected to trigger specific regenerative effects in vivo (3A) that could result in their direct pre-clinical/clinical translation (4). The matching between the elicited in vivo response (e.g. angiogenic, osteo-inductive or proliferative effect) and the ECM composition could lead to the selection of a set of factors (5A) capable to induce a desired regenerative process. Ultimately, the design of the resulting customized ECM would represent a new generation of instructive materials with enhanced performance for a predictable regeneration (6).

(B) The decellularization by apoptosis also offers the exciting but challenging possibility to investigate the properties of ECM from engineered/native tissues (1 b), through the removal of the cellular fraction without compromising the integrity of the ECM (2b). This may allow to better understand the role and function of the different ECMs in the absence of the cellular component. The instructive capacity of such acellular constructs can be evaluated upon in vivo implantation in suitable models (3b) and the observed regenerative effect can either lead to a direct pre-clinical/clinical translation of the generated ECM (4), or can be matched with the respective ECM composition (5b). A combinatorial analysis of this matching may allow the identification of new signals (5b) that critically drove the regenerative process. Ultimately, these factors could be implemented within the design of customized ECM (6), converging with the approach described above (Figure 5A) towards the generation of instructive materials that contain the necessary set of signals required for the repair of specific tissues/organs.

Examples

Materials and Methods

1. Cell isolation and culture

Bone marrow aspirates were obtained during routine orthopedic surgical procedures involving exposure of the iliac crest, after informed consent. Marrow aspirates (20 ml volumes) were harvested from healthy donors using a bone marrow biopsy needle inserted through the cortical bone; aspirates were immediately transferred into plastic tubes containing 15,000 IU heparin. After diluting the marrow aspirates with phosphate buffered saline (PBS) at a ratio of 1 :4, nucleated cells were isolated using a density gradient solution (Histopaquel , Sigma Chemical, Buchs, CH). Complete medium consisted of either Dulbecco's modified Eagle medium (DMEM) or a-minimum essential Medium (aMEM) with 10% fetal bovine serum, 1 % HEPES, 1 % Sodium pyruvate and 1 % of Penicillin- Streptomycin-Glutamin (100X) solution (all from Gibco). Nucleated cells were plated at a density of 3.10⁶cells/cm² in complete medium supplemented with 5 ng/ml of fibroblast growth factor-2 (R&D Systems) and cultured in a humidified 37C 5% C0₂ incubator. Medium was changed twice in a week. HMSCs were selected on the basis of adhesion and proliferation on the plastic substrate 1 week after seeding. This procedure is a standard protocol for the isolation of hMSC. Phenotypic analysis is typically performed later on (after a certain cell expansion) to confirm their MSC character.

2. Immortalization

Immortalization was performed by the use of a lentivirus Lenti-hTERT-GFP (LG508, Biogenova). Infection of hMSCs was performed 1 week after isolation and plating of the nucleated fraction from bone marrow aspirate. Virus was delivered at a MOI of 5 in complete medium supplemented with 8pg/ml_ of polybrene (Sigma Aldrich).

Typically, hMSCs are transduced using a MOI comprised between 1 and 20. The objective was to use relatively limited amounts of virus in order to not adversely affect the viability and differentiation potential of the cells, while having a high transduction efficiency. A MOI of 5 in hMSCs was shown to lead to a great transduction efficiency with a limited cell impairment. A MOI of 2.5 was also tested but led to a slightly lower transduction efficiency.

The success of immortalization was assessed by flow cytometry (GFP expression), telomerase activity measurement (TraP assay, Millipore, cat# S7700), through a senescence assay (β-galactosidase assay, Sigma Aldrich, cat# CS0030) and by following the population doublings (PD) of the cells. The formula (PD_(n/(n-1)=(log(N_n/N_n-1))/log2 was used for the calculations of the doublings between two passages, with n=passage and N=cell number. The cumulative population doubling levels (PDLs) is the sum of population doublings (PD).

3. Retrovirus production

The retro-vector carrying the modified caspase 9 and CD19 (iCasp9-ACD19) was kindly provided by Dr. Carlos Almeida Ramos (Baylor College of Medicine, Houston, Texas, USA; see references for Straathof et al. and Alameida Ramos et al. above). The retrovirus was produced after transfecting the phoenix ECO cell line ((ATCC product #SD3444; American Type Culture Collection, Manassas, VA, http://www.atcc.org) with the iCasp9-ACD19 vector. Virus containing supernatant was collected every 12 hours, passed through a 0.45 pm filter and conserved at -80°C.

The MSCV Luciferase PGK-hygro plasmid (Addgene plasmid 18782) was used for the production of retrovirus carrying the luciferase system (retro-Lucif). The virus production was performed using the same protocol as for the iCasp9-ACD19 retrovirus. Cells transduced with the retro-Lucif virus were selected by hygromycin B (Sigma, cat# H3274) resistance for a period of 2 weeks.

4. Cell sorting

For sorting experiments, a FACS-Vantage SE cell sorter (Becton Dickinson, Basel, Switzerland) was used. After immortalization with the lenti-hTERT-GFP, the immortalized population was sorted using the GFP marker. Following retroviral iDS transduction (iCasp9- ACD19), cells were sorted after labeling of the CD19 positive fraction using a human anti CD19-PerCP antibody (BD biosciences, cat# 561295). Cells were sorted exactly 1 month post-immortalization. Between immortalization and sorting, cells underwent 8 passages which correspond to approximately 30 PDLs.

5. Cell phenotyping

hMSCs phenotype was determined by cytofluometry analysis with fluorochrome-conjugated antibodies to human CD44 (cat# 559942), CD29 (cat# 555443), CD73 (cat# 560847), CD90 (cat# 559869), CD34 (cat# 555822), CD45 (cat# 555483), CD146 (cat# 550315), CD19 (cat# 561295), Epcam (cat# 347200) all from BD Pharmingen™, and E-cadherin (cat#FAB18381 P) from R&D Systems.

6. In-vitro differentiation assay

Adipogenic differentiation. hMSCs were seeded at 3000 cells/cm² and cultured for 1 week in DMEM complete medium without passage. During the following 2 weeks, cells were exposed to four differentiation cycles consisting in alternating ^'strong' adipogenic medium (Dexamethasone 10^"6 M, Indomethacin, Insulin, IBMX) for 3 days and ^'light' adipogenic medium (Insulin) for 1 day.

Osteogenic differentiation. hMSCs were seeded at 3000 cells/cm² and differentiated for 3 weeks in osteogenic medium (OM). Osteogenic medium consists in aMEM complete medium supplemented with 10 nM Dexamethasone, 0.1 mM L-ascorbic acid-2-phosphate and 10 mM β-glycerophosphate [Maniatopoulos et al; Cell Tissue Res. 1988 Nov;254(2):317-30].

Chondrogenic differentiation. Culture of hMSC was performed in 0.5 ml of DMEM containing 4.5 mg/ml D-Glucose, 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, 100mM HEPES buffer, 100 U/ml penicillin, 100 mg/ml streptomycin, and 0.29 mg/ml L-glutamine supplemented with 0.1 mM ascorbic acid 2-phosphate, 10 ng/ml TGFbl and 10^"7 M dexamethasone (Chondrogenic medium) as previously described [Jakob et al., Journal of Cellular Biochemistry 81 :368±377 (2001 ]. Cells (350,000) were centrifuged in 1.5 ml conical polypropylene tubes (Sarstedt, Nijmbrecht, Germany) to form spherical pellets. After 2 weeks of culture, pellets were processed histologically and for gene expression analysis. Endochondral construct generation. Endochondral constructs were generated as previously described [Scotti et al., Proc Natl Acad Sci U S A. 2010 April 20; 107(16): 7251-7256]. Briefly, hMSCs (500,000) were differentiated into chondrogenic lineages in transwell culture for 3 weeks in chondrogenic medium, followed by 2 weeks in a serum-free hypertrophic medium ( 50 nM thyroxine, 7.0 ^χ 10-3 M β-glycerophosphate, 10-8 M dexamethasone, and 2.5 x 10-4 M ascorbic acid).

7. Real-time PCR

Total RNA was extracted from cells using TRIzol (Invitrogen, Carlsbad, CA), treated with DNAse and retrotranscribed into cDNA, as previously described [Frank et al., Journal of Cellular Biochemistry 85:737-746 (2002)]. Real-time PCR was performed with the ABIPrism 77000 Sequence Detection System (Perkin Elmer/Applied Biosystem, Rotkreuz, Switzerland) and expression levels of genes of interest were normalized to GAPDH. Primers and probe sets of chondrogenic and osteogenic genes were used as previously described [Frank et al., ibid.]. Human PPARy (Hs00234592_m1 ) Retinoblastoma 1 , p53 (Hs01034249), p21 (Hs00355782_m1 ) primers and probe were provided by Applied Biosystem. Probe and primers for c-myc were ordered from Mycrosynth.

8. Stainings

Alizarin red, oil red-0 and Safranin-0 biochemical stainings were used to respectively assess the osteogenic, adipogenic and chondrogenic differentiation of the cells, as previously described [43, 44, 45]. Alcian blue (Sigma, cat# A5268) staining was also used to evidence a chondrogenic differentiation.

9. Tumorigenicity assay

One millions cells previously transduced with the retro-Lucif virus were mixed with Matrigel (BD Biosciences) 1 :1 PBS. The construct was implanted subcutaneously into the left flank of NOD/SCID mice. In vivo luminescence signal acquisition was performed using an ISIS luminometer system. Prior acquisition, mice received an injection of 300μΙ_ of D-Luciferin Potassium salt (Goldbiotechnology) at 15 mg/mL. Cells were tracked for a period of 3 months post-injection. The tumorigenicity assessment was assessed both by palpation and cell tracking using the luciferase system. The in vivo tracking of the cells represents a state-of- the-art and reliable method to assess the tumorigenic potentials of cells, while many study relies only on the palpation method. No histopathology could be performed as no tissue could be retrieved 6 months post-implantation. The fact that no tumors could be detected despite an extensive in vivo time in immuno-depleted animals is generally considered as a proof of safety. 10. In vivo Bone formation

In vivo ectopic bone formation was assessed as previously described [Braccini et al., Stem Cells. 2005;23:1066-1072]. Briefly, cells were resuspended in 30μΙ of Fibrinogen (20 mg/ml Baxter, Austria), quickly mixed with 30μΙ of Thrombin (6IU/ml Baxter, Austria) and loaded onto 35μg of bovine bone-derived granules (Bio-Oss^®, Geistlich Switzerland). The constructs were transferred for 15 minutes in a humidified incubator at 37 °C with 5% C02 to allow fibrin polymerization and implanted subcutaneously in CD-1 nu/nu nude mice (Charles River, Germany). After 8 or 12 weeks, the constructs were harvested, fixed overnight in 4% formalin, decalcified with a decalcification solution made of 7% EDTA (Sigma, cat# E5134) and 10% sucrose (Sigma, cat# S9378), paraffin embedded and

were obtained from different levels. Sections were stained with Masson's trichrome and observed microscopically to detect histological features bone tissue.

1 1 . Calcium deposition assay

Total calcium was measured using a Randox Kit (cat# CA 590) after rinsing and extracting the cell layer in 0.5N HCI as described by Jaiswal et al., 1997, J Cell Biochem 64:295-312. The amount of deposited calcium was expressed as pg per ml. of culture.

12. Immunomodulation assay (CFSE)

The immunomodulation of hMSCs was assessed as previously described [Gattinoni et al., Nat Med. 201 1 . 17(10):1290-7]. Briefly, hMSCs were seeded in 96-well plate at 20.000, 10.000, 5000 and 2000 cells/well. CD4+ lymphocytes were extracted from whole blood PBMCs by magnetic beads labeling (Miltenyi Biotec) and seeded on top of hMSCs at 0.1 *10^Λ6 cells per well after labeling with 2 μΜ CFSE (7 min at 37 °C). The CD4+cells were then stimulated or not with 1 pg/mL of phytohemagglutinin and the proliferation index was determined after 4 days by flow cytomtry and analyzed with FlowJo software (Treestar).

13. Immunocytofluorescence analysis of osteogenic proteins (Alkalyne Phosphatase,

Osteocalcin)

For the staining of osteogenic proteins, cells were trypsinized, counted and washed with FACS buffer. Cells were first stained for Alkalyne Phosphatase (ALP) in FACS buffer by 30 minutes incubation at 4°C with an anti ALP-APC conjugated antibody (R&D systems®, cat# FAB1448A). Cells were then washed and resuspended in a fixation/permeabilization buffer (R&D systems®, cat# FC007) for 10 minutes at RT. An anti OC-PE conjugated (R&D systems®, cat# IC1419P) was added for the intra-cellular staining of human osteocalcin (OC). Cells were washed and analyzed by flow cytometry. 14. Death induction

The B/B homodimerizer (Clontech, cat# 635060) was added at 50 nM to iCasp9-ACD19 transduced H SCs in culture medium to activate the apoptosis pathway through the dimerization of the modified caspase 9. Percentage of induced death was assessed 12 hours later by FACS analysis, after cell harvest and staining with Annexin V-APC (BD Biosciences, cat#550475) and Propidium Iodide (BD Biosciences, cat# 51-6621 1 E) in Annexin-V binding buffer (BD Biosciences, cat# 556454). Control cells were cultured in the same medium without exposure to the homodimerizer.

15. Immunofluorescence images

Following in vivo explantation, samples were fixed in 1 .5% paraformaldehyde (Sigma), decalcified with EDTA (Sigma) solution, embedded in optimal cutting temperature, and snap frozen in liquid nitrogen. Sections (5 pm thick) were incubated with the primary antibodies against CD31 (PECAM-1 ; BD Pharmingen). A secondary antibody labeled with Alexa Fluo 546 (Invitrogen) was used and DAPI was employed to stain nuclei. Fluorescence images were acquired using an Olympus BX-61 microscope.

Experimental Results

16. Immortalization of human bone marrow derived Mesenchymal stromal cells

Primary hMSCs (donor 196) were transduced with a lentivirus carrying the human telomerase gene (figure 1 ). The rate of transduction was assessed by flow cytometry through the enhanced GFP (eGFP) reporter gene. After transduction, 78% of the cell population expressed the transgene (figure 2). The eGFP positive fraction was sorted to increase the purity of the transduced population. The success of the immortalization was then assessed both by measuring the telomerase activity (figure 3) and by following the number of population doubling (PD, figure 4). The immortalized population (196hT) displayed a higher telomerase activity (4 to 5 fold higher) than their primary counterpart (primary 196), even after extensive proliferation (shown up to 100 doublings). Primary hMSCs stopped to proliferate after 35 PD while the immortalized hMSCs underwent more than 270 PD so far (figure 4). Primary cells became senescent after 35 PD (figure 5) while the 196hT population bypassed the senescence associated crisis. A stable hMSCs line was thus successfully generated.

17. Generation of iCasp9-ACD19 hMSCs line

After having undergone 170 PD, the 196hT line was transduced with the retrovirus carrying the inducible death system (iDS, figure 6). After one round of transduction, more than 50% of the cells expressed the CD19 surface markers (reporter gene). Cells were then labeled with an anti CD19-PerCP conjugated antibody for the sorting of the CD19-positive population. After sorting, more than 98% of the cells expressed this CD19 reporter protein (figure 7), resulting in the isolation of a stable hMSCs line carrying a death-inducible system (196hT- iDS). The functionality of this device was investigated by exposing the 196hT-iDS line to the Chemical Inducer of Dimerization (CID), thus activating the suicide system. Following overnight exposure to CID (figure 8), more than 99% of the cells were dead (Pl-positive) or apoptotic (Annexin-V-positive). Thus, an immortalized but death-inducible hMSCs line was successfully generated.

18. Selection of a potent clone based on osteogenic differentiation

The immortalization, the insertion of the iDS combined with the extensive number of doublings of the cells gave rise to a highly heterogeneous cell line with an impaired multi- differentiation potential (data not shown). A clonal selection based on osteogenic potential was thus performed in order to look for clones with potent differentiation potential, but also to standardize the properties of the cell line by obtaining a homogeneous clonal population. 30 single colonies were picked and re-seeded for expansion. Out of those 30 clones, 24 were successfully expanded and then differentiated for 3 weeks in osteogenic medium. A potent osteogenic clone was selected by alizarin red staining of the deposited mineralized matrix (figure 9). The clone #20 was qualitatively shown to secrete the most mineralized matrix. This immortalized hMSCs clone carrying the death device was then denominated as the Mesenchymal Sword Of Damocles (M-SOD) cell line.

19. Phenotypic analysis of primary 196 and generated cell lines

The phenotype of the different cell lines was assessed at different doublings and after each genetic modification (figure 10). All transformed cells expressed typical hMSCs markers (positivity for CD44, CD29, CD73, and CD90) while being negative for both the hematopoietic markers CD34 and CD45 and the epithelial markers EpCam and E-cadherin. The immortalized population was also shown to increasingly express the CD146 marker as soon as FGF-2 was removed from the supplemented medium (at 140 PD). The eGFP and CD19 expression is respectively linked to the immortalization and insertion of the iDS. The successive genetic modifications and extensive proliferation of the cell line did not alter their Mesenchymal phenotype.

20. Differentiation capacity of M-SOD cell line

The osteogenic, adipogenic and chondrogenic differentiation capacity of the M-SOD line was investigated in vitro.

The osteogenic differentiation capacity of M-SOD line was assessed after 3 weeks of culture in either osteogenic or complete medium. In the osteogenic culture condition, cells deposited a thick mineralized matrix as demonstrated by alizarin red staining (figure 1 1 ) and the high levels of calcium deposition (>5( g/ml_, figure 12). On the contrary, almost no mineral deposition was measured in complete medium condition (<0.5ug/ml_). A strong induction of key osteogenic genes was detected (22.7, 6.8, 22.6 and 6.4 fold for ALP, BMP-2, OC and BSP genes respectively) during culture in osteogenic conditions and confirmed at the protein level by immunocytochemistry for ALP and OC. Indeed, cells were shown to express ALP and OC proteins in temporal patterns (figure 14) similar to those described during the osteogenic differentiation of primary hMSCs. Thus, after 2 weeks of differentiation, more than 45% of the cells were positive for ALP (early osteogenic marker) and only 5% positive for OC (late osteogenic marker). After 3 weeks, the percentage of ALP positive cells decreased to 15% whereas OC positive cells increased to more than 20%. As a comparison, in complete medium conditions cells failed to express significant levels of either ALP and OC proteins (<5%), independently of time. The M-SOD cell line was thereby shown to be capable to strongly differentiate toward the osteogenic lineage.

The adipogenic differentiation of the M-SOD cells was investigated after 3 weeks of culture in either adipogenic or complete medium. In adipogenic culture conditions, cells were successfully differentiated into adipocytes as revealed by the positive Oil red-0 staining of the lipid droplets (figure 15). The overexpression of the adipogenic PPARy gene (15.8 fold, figure 16) confirmed the capacity of the M-SOD cell line to differentiate toward the adipogenic lineage.

The chondrogenic differentiation capacity of the M-SOD line was assessed by Alcian blue staining after pellet culture in chondrogenic medium (figure 17). M-SOD cells failed in generating a cartilaginous matrix. Nevertheless, the primary 196 donor, from which the M- SOD line was derived, was also shown to not be capable of chondrogenic differentiation (figure 17), strongly suggesting that M-SOD's absence of differentiation is not due to the successive genetic modifications and extensive doublings.

21 . Bone formation capacity of the M-SOD cell line

The bone formation capacity was assessed by mixing 1 million of M-SOD cells together with ceramic granules, in a fibrinogen/thrombin gel. The generated constructs were then implanted ectopically in nude mice for 8 weeks. Samples were retrieved, fixed and decalcified prior to sectioning and histological analysis. The cells secreted a dense collagen matrix and the formation of bone nodules was observed within the constructs, as demonstrated by Masson's trichrome staining (figure 18). GFP positive cells were observed around those nodules, strongly suggesting a human origin for this bone. Thus, the generated M-SOD cell line displays bone forming capacity despite an extensive proliferation and multiple genetic modifications. 22. Immunomodulation capacity

hMSCs are known to be able to modulate the immune response by regulating lymphocytes proliferation. To assess whether the M-SOD line triggers this important feature, a CFSE assay was performed using CD4+ lymphocytes. CD4+ cells were labeled with CFSE and seeded on top of M-SOD cells. Different ratios of the co-culture CD4+: MSCs were tested to better analyze a possible effect of M-SOD cells on CD4+ proliferation (1 :5, 1 :10 and 1 :20). The proliferation of CD4+ was measured 4 days post activation with phytohemagglutinin (PHA), by flow cytometry (figure 19, A). In the absence of PHA, CD4+ did not proliferate, independently of the presence or not of M-SOD cells. This implies that M-SOD cells alone do not trigger an immune reaction. In presence of PHA but in absence of M-SOD, CD4+ got activated and strongly proliferated (>80%), as shown by the 5 peaks of fluorescence observed, corresponding to 5 division cycles (figure 19, B). M-SOD cells were able to inhibit this CD4+ proliferation in a dose dependent manner. In particular, a 1 :5 CD4+/MSCs ratio better regulated CD4+ proliferation as only one peak (corresponding to only one division) could be observed (figure 19, B). The M-SOD cell line was thus shown to have immunomodulation properties by regulating the proliferation of activated CD4+ lymphocytes.

23. Death induction upon osteogenic and adipogenic differentiation

During cell differentiation, the chromatin is remodeled and can potentially lead to the silencing of the implemented iDS. Consequently, the functionality of the iDS system has to be demonstrated also during cell differentiation.

During osteogenic differentiation, cells were induced with CID after 1 , 2 or 3 weeks of culture in osteogenic medium. After an overnight exposure to CID, most of the cells were killed (>95%, figure 20) independently of their status of differentiation. Similar results were obtained by inducing M-SOD cells during adipogenic differentiation (figure 21 ). The iDS implemented within the M-SOD cell line remains highly efficient (>95% of cell death) independently of the adipogenic or osteogenic differentiation.

24. Oncogenes expression level and tumorigenicity assay

Cell immortalization, as any genetic modifications, can result in a malignant transformation of the cells through the impairment of the cell cycle regulation. Key proteins, such as p53, p21 , retinoblastoma-1 or c-myc, are known to play an important role as cell cycle regulator, keeping cell proliferation under control. The expression of those genes was shown to be often deregulated in many tumors. Therefore, the gene expression level of p53, p21 , retinoblastoma-1 and c-Myc was measured and directly compared to the ones of primary 196 (figure 22). No significant change in the expression was observed between the two populations (below 5 fold for each genes). The M-SOD cell line keeps expressing normal levels of cell cycle regulators gene. The tumorigenicity of the M-SOD cell line was then investigated in vivo to address possible safety concerns. One millions M-SOD cells, previously transduced with a luciferase reporter system, were injected subcutaneously in the flank of NOD/SCI D mice. Within few days, the luminescence intensity rapidly decreased suggesting the non-proliferation and dissemination of M-SOD cells (figure 23). After more than 50 days post-injection, neither luminescence nor tumor formation was detected indicating non-tumorigenicity in vivo. Mice remained healthy for at least 6 months.

25. Successful generation of 202-iDS cells

To compare the freeze & thaw (F&T) devitalization and the devitalization by apoptosis induction (figure 24), primary hMSCs (donor 202) carrying the iDS were generated. After transduction of the primary 202, cells were labeled with an anti CD19-PerCP conjugated antibody, to sort the positive population. More than 96% of the sorted population was shown to express the CD19 surface marker, thus carrying the iDS (figure 25). The resulting population was defined as 202-iDS and the primary hMSCs as 202. The two populations were then seeded and cultured on transwell plates to generate hypertrophic constructs.

26. Devitalization of hypertrophic constructs

After 5 weeks of transwell culture, the 202 and 202-iDS based constructs were devitalized by either F&T or iDS activation (+CID). Annexin-V (apoptotic cells) and PI (dead cells) staining measured the killing efficiency for each method (figure 26). Both methods led to a very high killing efficiency (>91 % using F&T and >93% using CID). Instead, in the non-devitalized constructs only 16% of the cells were dead. The devitalization by apoptosis induction was thus shown to be as efficient as the F&T technique.

27. Characterization of the generated hypertrophic constructs

The quality of the hypertrophic constructs (with or without devitalization) was assessed prior to the in vivo implantation. Both 202 and 202-iDS populations successfully generated a typical hypertrophic cartilaginous graft, consisting in a core rich in glycosaminoglycan (GAG, figure 27) and a mineralized outer ring (figure 28). The devitalization by F&T or by apoptosis induction (+CID) did not alter the quality of the matrix template, as qualitatively confirmed by the strong GAG staining (figure 27) and mineralization of the outer rim (figure 28). Quantitative real time RT-PCR (figure 29) showed a strong up-regulation of extra-cellular matrix (Collagen I), chondrogenic (Collagen II and SOX-9) and hypertrophic (Collagen X, MMP-13) markers, as compared to the expression levels of expanded cells. All constructs displayed very similar up-regulations of these genes, showing no deleterious effects due to cell transduction or devitalization step. Thus, prior to implantation the different constructs displayed all the characteristics of a mature hypertrophic cartilaginous graft. No qualitative differences were observed between the devitalized and the non-devitalized constructs. Nevertheless, those results do not actually exclude a possible loss of key ECM proteins during the devitalization procedure, especially regarding the F&T physical devitalization.

28. In vivo bone formation assessment

Hypertrophic constructs were implanted subcutaneously in the back of nude mice and retrieved 12 weeks later to investigate bone formation. After fixation and decalcification, samples were processed and stained by Masson's trichrome (figure 30). The non-devitalized samples (202 and 202-iDS) were completely remodeled into bone and the presence of marrow demonstrated the successful development into a fully mature organ. Instead, the F&T construct did not display evidence of any bone formation, the sample being only composed of an immature matrix. The devitalization by apoptosis induction (+CID) successfully led to the formation of perichondral bone surrounding the construct. The presence of bone marrow was also observed, though in lower quantity than what is observed in non-devitalized construct. While F&T did not allow bone formation, the iDS devitalization led to an intermediate stage of bone maturation when compared to the non-devitalized samples. Thus, the apoptosis devitalization method seems to be more supportive of bone generation than the F&T.

29. Vascularization of the graft

The vascularization of the graft is essential for the maturation of the cartilaginous template into a functional bone organ. A lack of vascularization is associated with hypoxia and prevents host cells penetration for tissue remodeling. During explantation of the constructs, clear differences were observed in terms of vascularization between the F&T and +CID devitalized samples (figure 31 ). The F&T constructs did not present external signs of vascularization while vessels and bloody tissue was observed in the apoptosis induced constructs. Those macroscopic observations were confirmed by CD31 immunofluorescence staining (figure 32). The F&T constructs were shown to be poorly vascularized, as evidenced by the detection of only few vessels specifically in the outer part of the graft. The devitalization by apoptosis led to a better engraftment of the samples as more CD31 cells were detected. Those vessels were also much more distributed inside the construct, from the outer part into the central core. Thus, the devitalization by apoptosis seems to be superior in terms of vascularization/ engraftment than the devitalization by F&T, likely due to a superior preservation of the composition, structure and function of the extracellular matrix.

Claims

A method to generate a mesenchymal stem cell preparation, particularly a human bone marrow-derived mesenchymal stem/stromal cell (hMSC) preparation, comprising the steps of

c. introducing, in an iDS transduction step, into said immortalized mesenchymal stem cells a second nucleic acid sequence encoding a fusion protein comprising i. a monomer of an apoptosis-inducing protein, wherein said apoptosis- inducing protein is able to induce apoptosis in said mesenchymal stem cell when said apoptosis-inducing protein is present in dimeric form, and ii. a protein able to bind to a dimerizing small molecule pharmaceutical drug, iii. wherein said fusion molecule dimerizes and trigger apoptosis in said

mesenchymal stem cell when said cell is exposed to said dimerizing small molecule pharmaceutical drug, and wherein

The method of claim 1 , wherein said first and/or second nucleic acid sequence is introduced by retroviral transduction.

The method of claim 1 , wherein said first nucleic acid sequence is introduced into said mesenchymal stem cells by lentiviral transduction.

The method according to any one of the previous claims, wherein subsequent to the immortalization step or subsequent to the iDS transduction step, a single clone is selected for further propagation.

The method according to any one of the previous claims, wherein said mammalian telomerase is a human telomerase.

The method according to any one of the previous claims, wherein said mammalian telomerase is the telomerase reverse transcriptase as defined in GenBank BAC1 1010 (Gene ID 7015).

7. The method according to any one of the previous claims, wherein said first and/or said second promoter sequence is constitutively expressed.

8. The method according to any one of the previous claims, wherein said first and/or said second promoter sequence is a cytomegalovirus (CMV) promoter or SC40 promoter, particularly the CMV immediate early promoter or SC40 early promoter.

9. The method according to any one of the previous claims, wherein said protein able to induce apoptosis is a caspase, particularly caspase 9 (Uniprot ID P5521 1 ).

10. The method according to any one of the previous claims, wherein said first and/or said second nucleic acid sequence comprises an expressed indicator transgene (by way of example: a fluorescent gene (EGFP) or a surface marker (CD19)) facilitating the selection of cells expressing said expressed indicator transgene.

1 1. The method according to any one of the previous claims, wherein said human bone marrow-derived mesenchymal stem/stromal cell preparation comprises at least 85%, 90% or 95% of cells that are CD73, CD90 and / or CD105 positive.

12. The method of claim 1 1 , wherein said human bone marrow-derived mesenchymal stem/stromal cell preparation comprises at least 85%, 90% or 95% of cells that are positive for CD73, CD90 and CD105.

13. The method of claim 1 1 or 12, wherein said human bone marrow-derived

mesenchymal stem/stromal cell preparation comprises at least 85%, 90% or 95% a. cell that are positive for CD44, CD29, and/or CD73;

b. cells that are negative for hematopoietic (CD34, CD45) and epithelial markers (Epcam, E-cadherin).

14. The method according to any of claims 10 to 13, wherein subsequent to the

immortalization step and/or the iDS transduction step, said cell preparation is submitted to a sorting step whereby cells expressing said indicator transgene are selected.

15. A mesenchymal stem cell, particularly a human bone marrow-derived mesenchymal stem/stromal cell (hMSC), comprising

a. a first transgene comprising a first nucleic acid sequence encoding a

mammalian telomerase under control of a first promoter sequence operable in said mesenchymal stem cell, and

b. a second transgene comprising a second nucleic acid sequence encoding a fusion protein comprising i. a monomer of an apoptosis-inducing protein, wherein said apoptosis- inducing protein is able to induce apoptosis in said mesenchymal stem cell when said apoptosis-inducing protein is present in dimeric form, and

ii. a protein able to bind to a dimerizing small molecule pharmaceutical drug, iii. wherein said fusion molecule dimerizes and trigger apoptosis in said

16. A mesenchymal stem cell according to claim 15, characterized in that

a. the cell is positive for CD73, CD90 and / or CD105;

b. the cell is positive for CD44, CD29, and/or CD73;

c. the cell is negative for CD34, CD45, Epcam, and E-cadherin;

d. said mammalian telomerase is a human telomerase, particularly telomerase reverse transcriptase as defined in GenBank BAC1 1010;

e. said protein able to induce apoptosis is a caspase, particularly caspase 9

(Uniprot ID P5521 1 ); and/or

f. said first and/or said second nucleic acid sequence comprises an expressed transgene facilitating the selection of cells expressing said expressed transgene (by way of example: a fluorescent gene (EGFP) or a surface marker (CD19)).

17. A cell preparation comprising more than 80% of mesenchymal stem cells according to claim 15 or 16.

18. A cell preparation generated by a method according to any of claims 1 to 14.

19. A cell preparation according to claim 17 or 18, wherein more than 70% of the cells of the preparation:

a. are positive for expression of CD73, CD90 and / or CD105; and or

b. undergo apoptosis upon exposure to said small molecule pharmaceutical drug; and/or

c. are positive for expression of said expressed indicator transgene.

20. A method for generating a tissue matrix, comprising the steps of: a. providing a support matrix, and

b. contacting said support matrix with a cell preparation derived by a method according to any of claim 1 to 14, or a cell preparation according to any of claims 17 to 19 in an ex-vivo cultivation step.

21. A method for generating a tissue matrix according to claim 20, further comprising the step of exposing said matrix to a dimerizing small molecule pharmaceutical drug capable of inducing apoptosis in said cell preparation.

22. The method of any one of steps 20 to 21 , wherein the ex-vivo cultivation step is performed under conditions suitable for osteogenic, adipogenic or chondrogenic differentiation.

23. The method according to any one of the previous claims 20 to 22, wherein said method is performed in a 3D perfusion bioreactor.

24. A tissue matrix obtainable by a method according to claim 20 or 22.

25. A devitalized tissue matrix obtainable by a method according to claim 20 to 22.

26. A devitalized tissue matrix according to claim 25, wherein the matrix is a mature hypertrophic cartilage template.

27. A devitalized tissue matrix according to claim 25, wherein the matrix is a bone tissue template.

28. A devitalized tissue matrix according to claim 25, wherein the matrix is a skin

template.