This study was performed to indentify the molecular mechanisms underlying the dysregulation of OPCs which is suggested to be the key factor responsible for remyelination failure in MS and TME (FRANKLIN and FFRENCH-CONSTANT, 2008;
KUHLMANN et al., 2008; PRINGPROA et al., 2010; ULRICH et al., 2008).
5.1 Transplantation of murine glial progenitor cell line BO-1 cell result in lack of remyelination and glial tumor formation
In the first study, a chemically induced demyelination model of the murine caudal cerebellar peduncle (CCP) was established and used for the investigation of the in vivo behavior of the transplanted murine OPC line BO-1. Ethidium bromide-induced demyelination of the CCP and transplantation of OPCs has been described so far only in rats (BLAKEMORE et al., 2003; FRANKLIN, 2002; LOUIS et al., 1992;
WOODRUFF and FRANKLIN, 1999). The present study revealed that the murine CCP also represents a suitable model to investigate the pathogenesis of chemically induced demyelination. The BO-1 cell line, which was previously obtained by spontaneous immortalization (PRINGPROA et al., 2008), showed only limited oligodendrocytic differentiation in response to RA or coculturing with astrocytes in vitro (PRINGPROA et al., 2010). Therefore one aim of the present study was to analyze the response of trans- planted BO-1 cells within the demyelinated CCP clarifying the question whether BO-1 cells may be induced to enforce oligodendrocytic differentiation in vivo when exposed to a stimulating environment.
Transplantation of BO-1 cells into the demyelinated CCP did not elicit remyelination as expected but an invasive and metastasizing murine giant cell glioblastoma was noticed at the injection site. The expression of NG2 and PDGFRα, by most of the tumor cells confirmed their neuroektodermal origin (NISHIYAMA et al., 2009;
RUSSEL, 1963) and visualization of eGFP verified that the tumor cells arose from the transplant. Control experiments of BO-1 cells into the normal CCP revealed the same malignant behavior of the transplant excluding the possibility that experimental
TMEV and STAT3 demyelination was causally related to the rapid tumor growth.
To test the remyelinating potential of transplanted BO-1 cells, the CCP of adult mice was demyelinated by injection of the DNA-intercalating substance ethidium bromide.
Injection of this reagent resulted in a focal demyelinated lesion at the injection site.
Though a lower dose was required compared to rats, the extent and time course of the demyelination process was comparable to that described in the rat (WOODRUFF and FRANKLIN, 1999). Injection of 0.025% ethidium bromide induced marginal clinical signs and distinct demyelination within the CCP. Higher doses were found to create lesions that extended into the adjacent lateral, spinal, or superior vestibular nucleus.
Transplantation of mitogen-expanded primary cells or immortalized cell lines generally bears the increased risk of tumor formation (EMERY et al., 1999). Studies in the peripheral nervous system have provided evidence for species-specific differences in the malignant potential of trans- planted cells (EMERY et al., 1999;
LANGFORD et al., 1988). In vivo, BO-1 cells maintained a high growth rate and formed tumors instead of differentiating into mature oligodendrocytes. Similarily, rat in contrast to human mitogen-expanded primary Schwann cells formed tumors following injection into the sciatic nerve of immunodeficient mice (EMERY et al., 1999). In contrast, immortalized rat OPCs, such as the CG4 cell line, have been frequently used for implantation into the demyelinated brain stem and spinal cord (FRANKLIN et al., 1996; FRANKLIN et al., 1995; OLBY and BLAKEMORE, 1996; TONTSCH et al., 1994; TOURBAH et al., 1997) causing extensive remyelination mediated by the transplanted cells. None of the studies described a tumorigenic growth. Previous studies have shown that BO-1 cells have a strong inherent proliferative capacity (PRINGPROA et al., 2008; PRINGPROA et al., 2010). Contrary to the rat CG4 cell line, BO-1 cells did neither require the addition of the growth-promoting B104 conditioned medium for maximum in vitro proliferation nor did they respond to thyroid hormones known to increase oligodendrocytic differentiation (PRINGPROA et al., 2008; PRINGPROA et al., 2010). However, proliferation of BO-1 cells was effectively modified in vitro by RA and fetal calf serum (FCS; PRINGPROA et al., 2008).
Moreover, CNPase expression in vitro was dramatically increased upon coculturing with astrocytes (PRINGPROA et al., 2008), which induced CNPase expression in 60-90% of BO-1 cells compared to about 30-40% in the absence of astrocytes while the percentage of MBP-positive BO-1 cells remained unaltered at about 5%
(PRINGPROA et al., 2008). Interestingly, the majority of transplanted BO-1 cells retained a highly proliferative NG2- and PDGFRα-positive OPC phenotype. In contrast, less than 10% of the cells expressed CNPase, MBP, or GFAP characteristic of oligodendroglial or astrocytic differentiation, respectively (MORENO-FLORES et al., 2003). This unexpected finding indicates that the demyelinated CNS environment did not only fail to promote oligodendrocytic development as expected but instead reduced the expression of CNPase as compared to the in vitro condition (PRINGPROA et al., 2008).
Ultrastructural examination revealed poorly differentiated neoplastic cells lacking morphological characteristics of differentiated astrocytes or oligodendrocytes. The observed ultrastructure of the neoplastic cells is in accordance with features described for a putative adult glial progenitor cell population in the central nervous system of monkeys and rats consisting of pale irregular nuclei with a thin layer of heterochromatin beneath the nuclear envelope, clumps of heterochromatin, irregular cellular shape, abundant mitochondria, rough endoplasmic reticulum, and lack of intermediate filaments (PETERS, 2004; PRINGPROA et al., 2010). Interestingly, intracytoplasmic bundles of glial intermediate filaments, which are generally considered to be a diagnostically relevant ultrastructural feature for astroglial tumors in humans (HADFIELD and SILVERBERG, 1972; LIBERSKI, 1998), were also lacking in a previous study detailing the morphology of cell transplantation-induced murine astrocytomas (KOPPEL et al., 1988). Therefore, the lack of intermediate filaments including GFAP immunoreactivity as observed in mouse astrocytomas (KOPPEL et al., 1988; PILKINGTON et al., 1985) may be either due to poor differentiation or species-specific properties. Based on these results, it can be concluded that the local environment of the CCP was insufficient to stop proliferation and induce oligodendroglial or astrocytic differentiation of the transplanted BO-1 cells.
TMEV and STAT3 Immortalization of cells being genetically engineered or naturally occurring does not
necessarily indicate neoplastic transformation. Accordingly, a number of different cell lines created by genetic manipulations and exhibiting high growth rates in vitro did not induce tumor formation after transplantation into the brain, spinal cord, or the peripheral nerve (JUNG et al., 1994; MORENO-FLORES et al., 2003; NOBRE et al., 2010; TIMMER et al., 2004; TIMMER et al., 2003). This implies that additional factors are defining the malignant potential of the transplanted cells. Although proliferation of BO-1 cells could be reduced in vitro, the percentage of BO-1 cells undergoing oligodendrocytic and astrocytic differentiation was limited upon treatment with RA or FCS, respectively (PRINGPROA et al., 2008). This is in striking contrast to studies using the rat CG4 cell line, where a strong expression of MBP after treatment with thyroid hormones was noted.
5.2 STAT3 represents a molecular switch inducing astroglial instead of oligodendroglial differentiateon of OPCs in TME
Gene Set Enrichment Analysis was performed to combine the data of gene expression microarray studies with the morphometrically assessed density of OPCs within the spinal cord from the same animals in order to search for transcriptional changes intimately associated with the occurrence of OPCs at the pathway level (HIERLMEIER et al., 2013; SUBRAMANIAN et al., 2005; ULRICH et al., 2010).
Accordingly, a cluster of 6 GO terms related to tyrosine phosphorylation and the JAK-STAT cascade represented the most interesting finding from the list of 26 GO terms significantly positively correlated to the density of OPCs. Notably, a recent meta-analysis identified the JAK-STAT cascade to be one out of 15 GO terms significantly positively correlated with demyelination in TME, EAE and a transgenic TNF-overexpressing mouse model of demyelination (HIERLMEIER et al., 2013), highlighting that JAK-STAT signaling is of general importance in demyelinating conditions. The JAK-STAT cascade is involved in a pleothora of physiological and pathological conditions including the inflammatory and immune response (DARNELL et al., 1994; RODRIGUEZ et al., 2006), and tumorigenesis (BRANTLEY and
BENVENISTE, 2008; HANSMANN et al., 2012). Basically, binding of extracellular ligands like cytokines, growth factors and hormones to their cell surface receptors leads to a recruitment of JAKs followed by activation of a context dependent set out of the seven different STATs by tyrosine phosphorylation. The activated STATs homo- or heterodimerize, translocate to the nucleus, bind to specific transcription factor binding motifs of the DNA and induce the expression of their target genes (HARRISON, 2012;
KISSELEVA et al., 2002; RAWLINGS et al., 2004).
Based on the initial identification of the JAK-STAT cascade as interesting candidate pathway, we focused our analysis on a manually curated list of genes involved in JAK-STAT signaling. The demonstration of up-regulated expression of STAT1, 2, 3, 4, and 6 as well as many upstream activators and downstream targets in TME allowed no unequivocal decision on which of the multiple STAT cascades is most important in the setting of OPC dysregulation. Over and above, the rate of transcription allows only a rough interpretation on the amount of protein present within the spinal cord (GERSTEIN et al., 2007), and basically gives no information concerning the phosphorylation dependent activation status of the JAK-STAT cascade (HARRISON, 2012). However, the different STAT proteins are activated by various upstream events with considerable specificity and are known to be involved in different processes (AKIRA, 1999). STAT1 and 2 are both participating in interferon signaling and the activation of the inflammatory response (DARNELL et al., 1994; LIU et al., 1998).
STAT4 and 6 are essential for the differentiation of T cells into the Th1- or Th2-phenotype and the activation of the adaptive immune response (LIU et al., 1998;
RODRIGUEZ et al., 2006). Remarkably, the STAT3 pathway is known to be a key regulator inducing astrocytic differentiation in neuroglial progenitor cells during embryogenesis (BONNI et al., 1997; CAO et al., 2010; HE et al., 2005; RAJAN and MCKAY, 1998), and represents a molecular hub for signaling pathways in glial tumors such as glioblastoma (BRANTLEY and BENVENISTE, 2008; HANSMANN et al., 2012). Furthermore, multiple genome-wide association studies implicate that the STAT3 gene seems to represent a MS susceptibility locus (JAKKULA et al., 2010;
LILL et al., 2012). Based on this knowledge, STAT3 was selected as the most
TMEV and STAT3 rewarding candidate molecule for further experimental analysis.
Immunohistology revealed a robustly elevated amount of STAT3- as well as p-STAT3-positive cells within the demyelinating lesions, mostly exhibiting an astrocytic morphology and co-expressing p-STAT3 and GFAP. Similarly, a robust and astrocyte-specific STAT3 antigen-expression is reported in normal human subjects as well as MS lesions (CANNELLA and RAINE, 2004). The high and moderate positive correlation of the density of p-STAT3-positive cells to the density of TMEV- and GFAP-positive cells, respectively suggests a causal linkage of the TMEV infection to STAT3-activation and the hyperplasia of astrocytes within the lesions. This is in agreement with the postulated key role of STAT3 signaling in the induction of astrogliosis after traumatic central nervous system injury (HERRMANN et al., 2008;
OKADA et al., 2006; WANNER et al., 2013; XIA et al., 2002). However, the latter studies also demonstrated that reactive astrocytosis is not always harmful, since it improves wound healing and functional recovery after traumatic CNS injuries, due to a restriction of the secondary inflammatory reaction (HERRMANN et al., 2008;
OKADA et al., 2006; WANNER et al., 2013). Since other studies suggest that reactive astrocytes mediate pro-inflammatory effects that increase disease severity in EAE and TME (BRAMBILLA et al., 2009; GERHAUSER et al., 2007; SOFRONIEW and VINTERS, 2010), and reactive astrogliosis can impede axon regeneration after CNS injury (FITCH and SILVER, 2008; SILVER and MILLER, 2004), the answer to the question whether reactive astrocytes are beneficial or harmful is most likely highly context dependent. A recent microarray study comparing reactive astrocytes in ischemia and neuroinflammation suggests the existence of different anti- and pro-inflammatory subtypes of reactive astrocytes, with Stat3 being ~2-fold up-regulated in both subtypes (ZAMANIAN et al., 2012).
The demonstration of spindloid cells co-expressing p-STAT3 and NG2 supports our GSEA results, suggesting that the JAK-STAT cascade is an important signaling pathway within intralesional OPCs. It seems reasonable that this active intranuclear p-STAT3 shifts the differentiation of the intralesional OPCs towards an astrocytic fate, since STAT3 is known to induce astrocytic differentiation in neuroglial progenitor cells
during embryogenesis (BONNI et al., 1997; CAO et al., 2010; HE et al., 2005; RAJAN and MCKAY, 1998). In agreement with this, a minor amount of intralesional NG2 and GFAP colocalization has been described in a previous TME study (ULRICH et al., 2008). It has to be mentioned that STAT3 is also an essential transcription factor in the differentiation of Th17-lymphocytes (CHEN and SHANNON, 2013; WANG et al., 2013). However, no prominent STAT3 expression was observed in the lymphocytes within the perivascular cuffs in the present study.
In order to verify the suggested molecular switch-like action of STAT3 on the differentiation of OPCs in vitro experiments using the glial progenitor cell line BO-1 were performed. Transplanted BO-1 cells exhibit a marked cytoplasmic STAT3 expression, whereas only few cells display an additional nuclear immunolabeling, indicative of an unaltered activation state similar to normal astrocytes (HANSMANN et al., 2012). Meteorin was used to activate STAT3 in a concentration reported to induce astrocytic differentiation in mouse embryonic forebrain neurospheres (LEE et al., 2010). In agreement with this, we could show increased astrocytic differentiation of the BO-1 cells when meteorin application was combined with astrocytic differentiation medium. In contrast, the concentration of STAT3 inhibitor VII, reported to inhibit STAT3 phosphorylation in cultured astrocytes (HASHIOKA et al., 2011), had to be reduced by 300-fold in order to decrease its anti-proliferative activity and prevent complete cytotoxicity in BO-1 cells in the present study. The used concentration of the STAT3 inhibitor VII combined with astrocytic differentiation medium was able to increase oligodendrocytic differentiation of BO-1 cells. These results supported our hypothesis that activated STAT3 shifts the differentiation of glial progenitors from an oligodendrocytic to an astrocytic fate. However, the interdependency of meteorin and STAT inhibitor VII upon the used medium suggests that other still unknown signaling mechanisms may influence OPC differentiation in addition.
5.3 Project overspanning interpretation
In summary, demonstration of a robust PDGFRα expression, an increased nuclear
TMEV and STAT3 p53 expression, and an unaltered STAT3 or activation state suggests that BO-1 cells
harbor molecular changes characteristic for secondary glioblastomas in humans (CALZOLARI and MALATESTA, 2010). Moveover, transplantation of the spontaneously immortalized murine glial cell line BO-1 did not only fail to remyelinate ethidium bromide-induced lesions of the murine CCP, they also lacked oligodendrocytic differentiation as identified by a dramatic reduction of CNPase expression compared to in vitro conditions. Moreover, BO-1 cells induced a giant cell glioblastoma in the demyelinated and non-demyelinated CCP. It is suggested that tumor formation follows a failure of molecular control mechanisms in BO-1 cells, responsible for a shift from proliferation to differentiation. Future studies should reveal whether an activation of the STAT3 signaling pathway can induce astrocytic differentiation and thereby limit the malignant potential of transplanted BO-1 cells.
Interestingly, TMEV-induced chronic progressive demyelinating leukomyelitis was accompanied by an increased population of intralesional NG2-positive OPCs and astrogliosis. Gene Set Enrichment Analysis (GSEA) suggested an association between OPCs and the JAK-STAT cascade. Immunohistochemistry revealed an increased intralesional amount of p-STAT3-positive cells mainly coexpressing GFAP followed by NG2 and CD107b positive cells. Activation or inhibition of STAT3 signaling in the glial progenitor cell line BO-1 cultured in astrocytic differentiation medium resulted in enhanced GFAP- or CNPase-expression, respectively.
Conclusively, STAT3 signaling represents an important factor shifting the differentiation of OPCs from an oligodendrocytic to an astrocytic fate thereby simultaneously inducing remyelination failure and glial scarring in chronic progressive demyelinating diseases like TME and possibly also MS.