Pancreatic differentiation (beta-like cells) of MSC

Cell therapy for replacement of dysfunctional or absent beta cells, secondary to auto-immune type 1 diabetes has been a major goal for stem cell research. Clinical islet transplantation allows already to restore beta-cell function after transplantation, (Marzorati, 2007), but the lack of organs to respond to the demand of numerous patients and also loss of islet function due to rejection and toxicity of

immunosuppressive regimen (Ryan, 2005), have motivated the search for alternative sources of beta cells.

MSC from different sources were used in experiments with the aim to differentiate them into beta-like or insulin secreting cells. Several groups isolated cells with a fibroblast-like morphology from the pancreas (Hao, 2006; Seaberg, 2004; Zhang, 2005b), from islets (Gershengorn, 2004; Zulewski, 2001) or exocrine tissue (Seeberger, 2006). They demonstrated that these cells had a mesenchymal phenotype (Seeberger, 2006). Gershengorn et al first derived fibroblast-like cells from human islets and demonstrated that these cells were able to proliferate extensively. Fibroblast-like cells did not express insulin, but vimentin and nestin. After expansion, serum deprivation reestablished epithelial morphology in these cells, and induced formation of islet-like clusters expressing insulin. The authors suggested that islet-cells underwent reversible epithelial to mesenchymal transition (EMT) to allow amplification and redifferentiation into beta-cells. Whether these cells are MSC was not established (Gershengorn, 2004). The possibility of islet cells to undergo EMT was addressed in subsequent studies in mice (Chase, 2007; Morton, 2007). Lineage tracing studies showed that islet-derived progenitor cells with MSC morphology may not derive from beta-cells through EMT, but represent a separate population (Morton, 2007). Chase et al showed in an elegant lineage tracing study that proliferating fibroblast-like cells isolated from pancreatic islets of mice were not derived from beta cells. They used the Cre-lox system in mice with Cre-recombinase under the Insulin 2 or the Pdx1 promoter and a floxed beta-geo-GFP construct inducing permanent GFP-expression in beta cells. Fibroblast-like cells did not express GFP and demonstrated a similar phenotype as MSC (Chase, 2007). Davani et al isolated fibroblast-like cells from human islets (Davani, 2007). These cells expressed CD73, CD90 and CD105 and were differentiated into osteoblasts, chondrocytes and adipocytes. MSC can thus be derived from human pancreatic islets, or from another view point, islet-derived progenitor cells possess a similar differentiation potential as MSC (Davani, 2007). Several other groups described similar cells isolated from the pancreas (Hao, 2006; Zhang, 2005b), ductal epithelium (Lin, 2006; Seeberger, 2006), exocrine fraction (Baertschiger, 2008) or cultured islets (Eberhardt, 2006;

Zulewski, 2001). The exact source of MSC has not been determined yet, but migration of MSC from the bone marrow into peripheral tissues was described (Sordi, 2005). Sordi et al described that MSC express different types of chemokine receptors and that cultured-islets release corresponding chemo-attractants. These results

suggest that the presence of MSC in the pancreas is secondary to migration from the bone marrow. Whether MSC reside within tissues since embryogenesis and constitute a ubiquitous niche or whether they migrate there from the bone marrow during post-natal life or in case of injury remains unknown.

MSC from pancreas sources were shown to express nestin and other pancreatic markers such as Pdx1 and Isl1 (Baertschiger, 2008; Eberhardt, 2006; Gallo, 2007b;

Lin, 2006; Zhang, 2005b). Nestin was thought to be a marker expressed only in neural progenitors, but this intermediate filament of the cytoskeleton was also shown in mesenchymal tissue within the pancreas (Street, 2004) and in bone marrow-derived MSC (Wautier, 2007).

MSC isolated from islets (Davani, 2007; Eberhardt, 2006), pancreatic tissues (Seeberger, 2006), umbilical cord blood (Gao, 2008), umbilical cord matrix (Chao, 2008b), adipose tissue (Timper, 2006) and bone marrow (Choi, 2005; Moriscot, 2005) were tested for their potential to differentiate into insulin secreting cells in vitro.

In islet-derived fibroblast-like cells insulin expression was induced with relatively simple methods, such as serum deprived medium (Gershengorn, 2004) or medium (CMRL-1066) without any growth factors (Davani, 2007).To compare differentiation potential of MSC from bone marrow and pancreatic / islet tissues, Mutskov et al studied histone modifications (H4 hyperacetylation and dimethylation of H3 lysine 4) displayed by the insulin gene in bone marrow-derived MSC, islet-derived MSC, and human islets (Mutskov, 2007). Indeed, they observed that in islets, the insulin gene was active and histones modified accordingly. In islet-derived MSC, chromatin modifications were also observed in regard to the insulin gene, as much as half of those seen in islets. In contrast, for bone marrow-derived MSC and HeLa cells no histone modification consistent with gene activation was observed (Mutskov, 2007).

They suggest that induction of insulin expression may be easier in MSC derived from pancreatic tissue and that bone marrow-derived MSC may need stronger signals or further genetic modifications to differentiate towards beta-like.

Addition of growth factors to culture medium or introduction of a plasmid coding for a beta-cell transcription factor or for insulin were tested to induce beta cell differentiation. Growth factors that were mostly used are activin A, HGF, exendin-4 (a glucagon-like-peptide-1 analog), IGF-1, keratinocyte growth factor (KGF), betacellulin, EGF, TGFbeta1, pentagastrin, but also other culture supplements such as retinoic acid, nicotinamide, B27 and N2 supplements (Abraham, 2002; Eberhardt, 2006; Gao, 2008; Seeberger, 2006; Timper, 2006; Zhang, 2005b). These factors

were applied either in a sequential manner, to repeat an embryonic-like differentiation process, or all at the same time. Analysis of the expression of beta-like transcription factors, such as Pdx1, Foxa2, Isl1, Ngn3, Nkx2.2, Nkx6.1, NeuroD, Pax4 and Pax6, and other beta-cell markers, such as insulin, C-peptide, glucokinase, GLUT2, by RT-PCR and at the protein level was used to show endocrine differentiation (Abraham, 2002; Eberhardt, 2006; Gao, 2008; Li, 2007c; Moriscot, 2005; Seeberger, 2006; Sun, 2006; Timper, 2006; Zhang, 2005b). It has been suggested that in pancreas-derived progenitors, expression of these transcription factors may be easier, as chromatin is in an open, active state, similar to the insulin gene (Mutskov, 2007).

Similar to protocols developed for liver differentiation using damaged hepatocyte extracts, Choi et al used pancreatic extracts from regenerating rat pancreatic tissue, 2 days after partial pancreatectomy and differentiated bone marrow-derived MSC towards endocrine cells, expressing most markers and releasing insulin in response to glucose (Choi, 2005). The authors suggested that regenerating pancreatic tissue contains factors with endocrine differentiation potentials (Choi, 2005).

Much effort has been invested in improving differentiation protocols, adding new factors such as conophylline (Hisanaga, 2008), an activin A agonist which showed to increase beta-cells during development (Ogata, 2004), or applying protocols developed for ESC differentiation (D'Amour, 2006. Three dimensional culture systems, to form pseudo-islet structures, also improved differentiation consistently.

Chao et al cultured umbilical cord (Wharton jelly)-derived MSC in neuronal conditioned medium and allowed cluster formation. These clusters expressed beta-cell transcription factors and demonstrated glucose-dependent insulin secretion.

Transplantation of differentiated clusters reversed streptozotocin induced diabetes in rats, whereas undifferentiated cells did not influence glycemic control (Chao, 2008b).

Hanging drop cultures of MSC, or addition of extra-cellular matrix and fibronectin in 3 dimensional culture systems also allowed beta-cell marker expression in human islet-derived (Davani, 2007), bone marrow-islet-derived (Chang, 2008) and UCB-islet-derived (Gao, 2008) MSC, whereas no insulin expression was observed when cells were cultured in mono-layers. Three dimensional structures and extra-cellular matrix, either supplemented in cultures or possibly secreted by cells themselves mediate signals to promote differentiation. These culture systems may reproduce interactions and mechanisms occurring in islets structures (Bosco, 2000; Bosco, 1989).

Introduction of beta-cell transcription factors into MSC by transfection and adenoviral or lentiviral-mediated transduction were proposed to force differentiation into beta-like

cells. Genetic modifications of bone marrow-derived MSC induced a partial beta-cell phenotype with in vitro and in vivo function. Introducing Pdx1 (also known as Ipf-1) into MSC by adenoviral transduction was inefficient, since bone marrow-derived MSC expressed only low levels of insulin and few beta-cell transcription (Foxa2, Isl1 and Pax4) (Moriscot, 2005). No insulin was observed at the protein level. Transduction of early pancreatic transcription factors, such as Foxa2 or HLXB9 did not increase expression of insulin or other transcription factors (Moriscot, 2005). Using human Pdx1 and adenovirus-mediated transfection in human bone marrow-derived MSC, Li et al showed expression of multiple beta-specific genes i.e. Ngn3, insulin, glucokinase, GLUT2, and glucagon. These cells were able to secrete insulin in response to glucose and IBMX (Li, 2007c). They also decreased glycemia to levels slightly above normo-glycemia, when transplanted under the kidney capsule of diabetic mice, demonstrating their function in vivo (Li, 2007c). Stable insertion of Pdx1 by retroviral gene transduction into the genome of bone marrow-derived MSC generated similar results with expression of multiple beta-cell genes, insulin secretion in response to glucose in vitro and partial function in vivo with a moderately reduced glycemia compared to non-transplanted animals (Karnieli, 2007). Whether transplanted cells express adenoviral or retroviral proteins on their surface after transduction or if random DNA integration of the gene after retroviral/lentiviral transduction were not investigated. Both points may hamper these approaches as they may favor rejection or induce cellular degeneration and tumor formation after transplantation in potential clinical application.

In vivo beta cell function is the ultimate aim to reach in this research field, before possible clinical application. Transplantation of bone marrow cells or undifferentiated MSC isolated from the bone marrow seem hardly to differentiate into functional beta-cells or islets (Chao, 2008b; Choi, 2003; Wu, 2007), as hyperglycemia alone might be a weak differentiation signal. However, after differentiation induction in vitro, with growth factors (Wu, 2007), genetic modification (Karnieli, 2007) or three-dimensional structure formation (Chao, 2008b; Chen, 2004; Davani, 2007), MSC were able to express insulin after transplantation and partially restore normo-glycemia.

Chronic inflammation and tissue damage to the pancreas favors migration of intravascular administered MSC (Lee, 2006). Lee et al injected repeatedly low doses of streptozotocin to mice and transplanted MSC which homed to the pancreas and kidneys, both sites where toxicity of streptozotocin was demonstrated (Lee, 2006).

Compared to control mice, mice transplanted with MSC demonstrated better

glycemic profile with larger pancreatic islets containing some human derived, differentiated beta-cells, higher mouse beta-cell numbers and fewer kidney lesions (Lee, 2006). Another study showed that MSC and bone marrow infusion in a mouse model of diabetes reversed hyperglycemia by decreasing inflammation and stimulating beta-cell regeneration (Urban, 2008).

MSC are currently under investigation for possible clinical application in co-transplantation with islets to improve co-transplantation outcome. Two characteristics of MSC seem to foster their use in this setting. First their immunomodulatory properties (Abdi, 2008; Le Blanc, 2007a); second their trophic effect and ability to sustain revascularization processes. Co-culture of islets with MSC and/or endothelial cells improve islet survival during culture, insulin secretion and revascularization processes in vitro (Chao, 2008a; Johansson, 2008).

In summary, MSC from various sources including islet-derived and pancreas-derived progenitors can express beta-cell markers and induce expression of insulin. Some epigenetic evidence shows that progenitors derived from the pancreas are easier to differentiate in vitro. Further quantification studies and comparison with islets are required to predict if differentiated cells will show a strong enough function to be applicable in clinical therapies