Grey matter OPCs are less mature and less sensitive to IFNγ than white matter OPCs: consequences for remyelination

Multiple sclerosis (MS) is a chronic inflammatory disease characterized by the formation of demyelinated lesions in the central nervous system. At later stages of the disease repair in the form of remyelination often fails, which leads to axonal degeneration and neurological disability. For the regeneration of myelin, oligodendrocyte progenitor cells (OPCs) have to migrate, proliferate and differentiate into remyelinating oligodendrocytes. Remyelination occurs faster and is more extensive in grey matter (GM) lesions than in white matter (WM) lesions. Here, we examined differences in neonatal OPCs from GM (gmOPCs) and WM (wmOPCs), both intrinsically and in response to environmental (injury) signals. We show that gmOPCs are less mature than wmOPCs, both on morphological and on gene-expression level. Additionally, gmOPCs proliferate more and differentiate slower than wmOPCs. When exposed to astrocyte-secreted signals wmOPC, but not gmOPC, migration decreases. In addition, wmOPCs are more sensitive to the detrimental effects of IFNγ treatment on proliferation, differentiation, and process arborisation, which is potentiated by TNFα. Our results demonstrate that OPCs from GM and WM differ both intrinsically and in response to their environment, which may contribute to the difference in remyelination efficiency between GM and WM MS lesions.

GmOPCs proliferate more and differentiate slower than wmOPCs. Upon demyelination one of the first events is the migration of activated adjacent OPCs to the lesioned area. To assess whether gmOPCs and wmOPCs differ in their migratory capacity, gmOPCs and wmOPCs were cultured on a porous membrane, and cellular migration towards a PDGF-AA gradient was examined. The percentage of gmOPCs that have migrated in 4 hours across the transwell membrane was similar to the percentage of migrated wmOPCs (Fig. 3a,b, respectively 10.7 ± 1.7% and 8.8 ± 1.2%, p = 0.146). In addition, OPCs have to proliferate to obtain sufficient numbers for successful remyelination. After a 48-hour exposure to the mitogens PDGF-AA and FGF-2, the percentage of cells positive for the proliferation marker ki67 was higher in gmOPCs than in wmOPCs (Fig. 3c, 36.8 ± 5.6% and 28.3 ± 3.2%, p = 0.048). The final step in remyelination is the differentiation of OPCs towards mature myelinating OLGs. Fluorescent imaging of GalCer/sulfatide by R-mAb showed that wmOPCs that maturated into wmOLGs were larger than gmOPCs that maturated into gmOLGs (Fig. 3e) corroborating a recent finding that OLGs show regional heterogeneity in morphology 37 . Immunofluorescent labelling of MBP, a marker for mature OLGs, showed that significantly more wmOPCs expressed MBP after 3 days of differentiation than gmOPCs (Fig. 3f,h, respectively 24.4 ± 3.0%and 15.7 ± 2.5%, p = 0.0003). However, after 6 days of differentiation the percentage of MBP-positive cells of gmOPCs and wmOPCs was similar (Fig. 3g,h, respectively 38.8 ± 4.5% and 37.8 ± 5.2%, p = 0.862). A read-out parameter for 'myelination' in OLG monocultures is the number of cells that elaborate MBP-positive myelin membranes of total MBP-positive cells. GmOLGs and wmOLGs hardly Figure 1. GmOPCs are morphologically less mature than wmOPCs. Oligodendrocyte progenitor cells (OPCs) isolated from the cortex (gmOPCs) and non-cortex (wmOPCs) of neonatal rat forebrains were cultured in the presence of PDGF-AA and FGF-2 for 48 hours. (a) Schematic representation of dissected areas of neonatal rat forebrains to obtain gmOPCs and wmOPCs (b) OPCs stained with the OPC cell surface marker antibody A2B5. Representative images are shown. (c-h) Analysis of the morphology of (b) using Sholl analysis of gmOPCs and wmOPCs of the same batch. The process length (c), the number of branch points (d), the number of processes that intersect with the concentric circles of the Sholl analysis as a function of the distance from the soma (f), maximum intersections (g), distance of the soma with the maximum number of branch points (h) and area under the curve (i) are shown. A representative image of the Sholl analysis is shown in (e). Note that while the process length is similar, wmOPCs have more branch points, a larger maximum of intersections and total area under the Sholl curve, indicating a more complex branched phenotype than gmOPCs. Bars represent mean process length (c), branch points (d), maximum intersections (g), distance of the soma with the maximum number of branch points (h) and area under the curve (i) of five independent experiments (24-26 cells analysed per independent experiment). Error bars show the standard error of the mean. Statistical analyses were performed using a paired two-sided t-test (*p < 0.05, **p < 0.01). Scale bar is 50 µm.
SCieNTifiC REPORtS | (2018) 8:2113 | DOI:10.1038/s41598-018-19934-6 differed in their ability to form myelin membranes in vitro (Fig. 3g,i, respectively 39.8 ± 10.6% and 39.3 ± 9.4%, p = 0.926). As the percentage of MBP-positive cells reaches a maximum at day 6, these findings indicate that the differentiation of wmOPCs was accelerated in vitro. Hence, in vitro gmOPCs proliferated more, while wmOPCs differentiated faster and elaborated more extensive process networks. Next to intrinsic differences in functional behaviour, a distinct response of gmOPCs and wmOPCs towards micro-environmental signals may also contribute to differences in (re)myelination efficiency.
WmOPCs migrate less in response to astrocyte conditioned medium than gmOPCs. Astrocytes are important regulators of OPC behaviour 38,39 . To examine whether gmOPCs and wmOPCs respond differently to astrocyte-derived factors we exposed gmOPCs and wmOPCs to astrocyte conditioned medium (ACM) and determined the effect on OPC migration, proliferation and differentiation. To this end, non-conditioned medium (NCM) and ACM were added to OPCs for the duration of the experiment. Exposing OPCs for 24 hours to ACM impairs wmOPC, but not gmOPC migration [ Fig. 4a, respectively 0.76 ± 0.4 (p = 0.003) and 1.86 ± 0.97 (p = 0.427) fold change], while proliferation of gmOPCs and wmOPCs was hardly affected upon ACM exposure Hence, wmOPCs were more receptive to astrocyte secreted signals, which alter functional endpoints relevant for myelination, i.e., migration and differentiation. In MS lesions, other factors like the pro-inflammatory cytokines TNFα and IFNγ may influence remyelination capacity of OPCs 24,[40][41][42][43][44][45] . Also, pro-inflammatory cytokines seem to play a role in the pathology of rodent models of MS, including experimental autoimmune encephalomyelitis and cuprizone-induced demyelination [46][47][48] . Therefore, we next examined the effect of TNFα and IFNγ on gmOPC and wmOPC morphology and behaviour in vitro.
IFNγ increases the process length of gmOPCs and wmOPCs and reduces the number of branch points in wmOPCs. Inflammation is a hallmark of MS, and pro-inflammatory cytokines TNFα and IFNγ have been demonstrated to play a role in the disease 45,49-51 . OPCs in an MS lesion environment have been exposed to these cytokines, which could result in an altered remyelination capacity. We asked whether exposure of OPCs GmOPCs are less mature than wmOPCs on gene expression level. Oligodendrocyte progenitor cells (OPCs) isolated from the cortex (gmOPCs) and non-cortex (wmOPCs) of neonatal rat forebrains were cultured in the presence of PDGF-AA and FGF-2 for 48 hours. OPC were subjected to qPCR analysis of markers that (i) inhibit OPC differentiation (a, immature markers), (ii) are associated with myelination (b, mature markers, in ascending order of maturity) and (iii) are enriched in differentiation committed OPCs (c, COP markers). Hmbs was used as reference gene; the reference gene Eef1a1 showed similar results (data not shown). Note that the mRNA expression levels of the more mature OPC markers are increased in wmOPCs compared to gmOPCs. Bars represent mean expression levels relative to gmOPCs, which were set at 1 for each independent experiment (horizontal line). Error bars show the standard error of the mean. Statistical analyses were performed using a one-sample t-test (*p < 0.05, **p < 0.01, ***p < 0.001, n = 4). Remarkably, the effect of IFNγ on wmOPCs was potentiated upon combined treatment with TNFα (Fig. 5d, 0.30 ± 0.04 fold change, p = 0.004), while exposure to TNFα was seemingly ineffective both in gmOPCs and Note the higher percentage of proliferating gmOPCs compared to wmOPCs. (e-i) OPC were differentiated for 3 (e,f, and h) and 6 days (g-i) and incubated with either (e) R-mAb, recognizing GalCer/sulfatide, or (f-i) double stained for MBP (red), a mature marker of oligodendrocytes (OLGs) and Olig2 (green), OLG lineage marker. Representative images are shown in (e,f) and (g); quantitative analyses of the number of MBP-positive OLGs of total Olig2-positive cells in (h) (n = 8 for 3 days, n = 10 for 6 days, at least 150 cells analysed per independent experiment) and the number of MBP-positive cells that elaborate myelin membranes in (I) (n = 10, 6 days). Note that after 3 days of differentiation wmOLGs are larger and morphologically more complex than gmOPCs. In addition, wmOPCs show an accelerated differentiation, while the number of MBP-positive cells bearing myelin membranes at day 6 is similar. Bars represent means. Error bars show the standard error of the mean. Statistical analyses were performed using a paired two-sided t-test (*p < 0.05, ***p < 0.001). Scale bar is 50 µm. Sholl analysis further revealed that IFNγ treatment reduced the maximum number of process intersections in wmOPCs compared to untreated wmOPCs (Fig. 6b,e, respectively 3.49 ± 0.29 and 5.16 ± 0.42, p = 0.004), but not in gmOPCs (Fig. 6a,c, respectively 3.37 ± 0.29 and 3.84 ± 0.34, p = 0.786). However, in gmOPCs the maximum number of intersections shifted towards a higher distance from the soma upon exposure to IFNγ (Fig. 6a,d, respectively 66.0 ± 4.6 and 30.6 ± 3.3, p = 0.0003), which was not evident in wmOPCs (Fig. 6b,f, respectively 59.0 ± 13.2 and 39.6 ± 1.5, p = 0.152). Remarkably, combined treatment of IFNγ with TNFα counteracted the effect of IFNγ in gmOPCs, i.e., the distance of the soma with the maximum number of intersections was similar to untreated control and TNFα-treated gmOPCs (Fig. 6a,d, 30.6 ± 3.3 and 36.0 ± 3.5, p = 0.270). Hence, these findings indicate that IFNγ treatment reduces OPC process arborisation, i.e., OPCs appear morphologically less mature upon IFNγ treatment, which was more pronounced in wmOPCs than gmOPCs.

WmOPCs are more sensitive to TNFα-and IFNγ-mediated inhibition of proliferation than gmOPCs.
To examine the effect of pro-inflammatory cytokines on cell behaviour aspects that are relevant to OPC recruitment, we next examined the effect of TNFα and IFNγ on gmOPC and wmOPC migration and proliferation compared to their respective untreated control OPCs. Upon exposure of TNFα, IFNγ or a combination of TNFα and IFNγ, the number of migrating cells was similar in gmOPCs [ Fig. 7b, respectively, 0.81 ± 0.09 (p = 0.100), 0.90 ± 0.11 (p = 0.418) and 0.86 ± 0.12 (p = 0.334) fold change] and wmOPCs [ Fig. 7b, respectively, 0.98 ± 0.12 (p = 0.904), 1.12 ± 0.08 (p = 0.202) and 1.12 ± 0.08 (p = 0.202) fold change]. Exposure to TNFα resulted in a decrease in proliferation in wmOPCs (Fig. 7a,c, 0.84 ± 0.04 fold change, p = 0.021), but not in gmOPCs (Fig. 7a,c, 1.01 ± 0.12 fold change, p = 0.945). Similarly, exposure to IFNγ significantly decreased wmOPC, but not gmOPC proliferation [ Fig. 7a,c, respectively 0.46 ± 0.12 (p = 0.017) and 0.74 ± 0.08 (p = 0.051) fold change]. Combined treatment of TNFα and IFNγ synergized in wmOPCs, resulting in a further decrease of proliferation (Fig. 7a,c, 0.35 ± 0.16 fold change, p = 0.028). GmOPC proliferation also decreased upon exposure to both TNFα and IFNγ (Fig. 7a,c, 0.60 ± 0.07 fold change, p = 0.013). Hence, these data indicate that wmOPCs were more sensitive to TNFα-and IFNγ-mediated inhibition of proliferation than gmOPCs. To examine whether the increased sensitivity of wmOPCs was mediated via an increased expression of the TNFα and/or IFNγ receptor, qPCR analysis was performed. The mRNA level of the IFNγ receptor Ifngr1, but not Ifngr2, was higher in wmOPCs than in gmOPCs IFNγ delays wmOPC, but not gmOPC differentiation. Upon toxin-induced demyelination, and likely also in MS lesions, OPCs are only transiently exposed to pro-inflammatory cytokines. To mimic the effect of this transient exposure to inflammatory signals, OPCs were treated with TNFα, IFNγ or a combination of TNFα and IFNγ for 48 hours, after which OPCs were allowed to differentiate in the absence of cytokines. Upon 3 days of differentiation, a brief exposure to IFNγ at the OPC stage decreased the percentage of MBP-positive . WmOPCs migrate less in response to astrocyte secreted factors than gmOPCs. Oligodendrocyte progenitor cells (OPCs) isolated from the cortex (gmOPCs) and non-cortex (wmOPCs) of neonatal rat forebrains were treated with non-conditioned medium (NCM) or cultured in the presence of astrocyte conditioned medium (ACM) at the indicated time points for the duration of the experiment. (a) OPC migration towards a 10 ng/ml PDGF-AA gradient (4 hours) was determined using a transwell assay (n = 5). Note that exposure to ACM decreased the migration of wmOPCs compared to NCM treatment, while gmOPC migration tends to increase upon ACM exposure. (b) Following 24 hours in culture, OPCs were exposed to NCM and ACM for 24 hours in the presence of PDGF-AA and FGF-2. OPC proliferation was determined by immunocytochemistry for the proliferation marker ki67 (n = 7). (c-e) OPCs were differentiated in NCM or ACM for 3 (c) and 6 days (d,e) and subjected to a double staining for MBP (red), a mature marker of oligodendrocytes (OLGs) and Olig2 (green), an oligodendrocyte (OLG) lineage marker (n = 5). Note that ACM increased differentiation (c,d), while myelin membrane formation is hardly affected (e). Bars represent mean relative to their respective NCM-treated control, which was set at 1 for each independent experiment (horizontal line). Error bars show the standard error of the mean. Statistical analyses were performed using a one-sample t-test (*p < 0.05, **p < 0.01) to test for differences between treatments and their respective control and an unpaired two-sided t-test was used to test whether the response to ACM differed between gmOPCs and wmOPCs (not significant).  Fig. 8a,b, respectively 0.19 ± 0.14 (p = 0.009) and 0.10 ± 0.06 (p = 0.001) fold change]. At 6 days of differentiation the decrease in gmOPC differentiation upon treatment with IFNγ was diminished (Fig. 8a,c, 0.94 ± 0.24 fold change, p = 0.818), while a slight but not significant decrease in wmOPC differentiation was still apparent upon TNFα and IFNγ exposure (Fig. 8a,c, 0.47 ± 0.17 fold change, p = 0.055). The percentage of MBP-positive OLGs that form myelin membranes at 6 days of differentiation was hardly affected when gmOPCs and wmOPCs were transiently exposed to the pro-inflammatory cytokines [ Fig. 8a,d, respectively 1.95 ± 0.64 (p = 0.232) and 1.07 ± 0.06 (p = 0.351) fold change upon TNFα exposure; 1.46 ± 0.96 (p = 0.663) and 1.24 ± 0.27 (p = 0.432) fold change upon IFNγ exposure; 1.40 ± 0.86 (p = 0.675) and 0.41 ± 0.24 (p = 0.090) fold change upon combined TNFα and IFNγ exposure]. Note that after 6 days of differentiation wmOLGs produced more elaborate myelin membranes than gmOLGs, consistent with the more elaborated network at day 3 of differentiation ( Fig. 8a cf Fig. 3e). Hence, brief exposure to IFNγ at the OPC stage delays wmOPC, but not gmOPC differentiation in vitro and transient exposure to a combination of TNFα and IFNγ may perturb wmOPC differentiation.

Discussion
Remyelination at physiological conditions and in MS is more extensive in GM lesions than in lesions of the WM 7,[9][10][11] . Here, we aimed to unravel whether inherent differences in gmOPC and wmOPC behaviour, including their response to environmental (injury) signals, contribute to regional differences in remyelination efficiency. Our in vitro findings -based on morphology, proliferation and migration capacity, differentiation kinetics and expression of myelination-associated genes-revealed that neonatal gmOPCs are less mature than neonatal wmOPCs. In addition, wmOPCs were less migratory upon addition of astrocyte secreted factors than gmOPCs, Note that IFNγ increases the process length of gmOPCs and wmOPCs (c) and reduces the number of branch points in wmOPCs, but not gmOPCs (d). When IFNγ is combined with TNFα the number of branch points is decreased in either OPC. Bars represent mean relative to their respective untreated control, which was set at 1 for each independent experiment (horizontal line). Grey bars represent gmOPCs, white bars represent wmOPCs. Error bars show the standard error of the mean. Statistical analyses were performed using column statistics with a one-sample t-test (*p < 0.05, **p < 0.01,***p < 0.001) to test for differences between treatments and their respective control and a one-way ANOVA with a Šidák post-test was used to test whether the response to TNFα, IFNγ and TNFα and IFNγ combined differed between gmOPCs and wmOPCs (not significant). Scale bar is 50 µm. and wmOPCs were also more sensitive to IFNγ-mediated inhibition of proliferation and differentiation than gmOPCs, an effect that was potentiated by TNFα. Given that OPCs revert to a more immature stage upon demyelination 52 , gmOPCs may have evolved to be better equipped for remyelination than wmOPCs, i.e., gmOPCs are more proliferative, less mature, and less responsive to astrocyte-derived factors that affect recruitment and less susceptible to inflammatory mediators than wmOPCs. The more matured stage of wmOPCs may provide an advantage in developmental myelination and myelin remodelling.
Remarkably, most in vitro studies have been performed with GM (cortical) OPCs, whereas remyelination and myelination in vivo is usually examined in WM areas. Our detailed in vitro comparison between neonatal gmOPCs and wmOPCs allowed us to characterize inherent differences between and responsiveness of regional OPCs without the interference of spatial and interacting environmental cues. In vitro, gmOPCs were less branched and had a higher capacity to proliferate in response to PDGF-AA and FGF-2 than wmOPCs. Indeed, relatively more OPCs are present in GM MS lesions, while endogenous remyelination in WM MS lesions is hampered by the recruitment of OPCs to the lesion site 7,10,53 . The in vitro differentiation of wmOPCs was accelerated compared to gmOPC differentiation. Previous fate mapping studies during CNS development showed that wmOPCs produce more mature myelinating OLGs, while gmOPCs remain immature 18 . Similarly, gmOPC, but not wmOPC maturation is retained when transplanted to either a healthy GM or WM area of the adult mouse brain 23 . Hence, while gmOPCs differentiate in vitro eventually to a similar extent as wmOPCs, in vivo gmOPCs remain immature irrespective of their environment. The relative maturity of wmOPCs, as evident by a more complex morphology and an increase in the level of genes that are associated with OPC maturation may underlie the accelerated maturation of wmOPCs in vitro. In the adult brain a continuum from OPCs to mature OLGs exists, and using a selection of genes, including genes that mark the so-called COPs, i.e., differentiation committed more mature OPCs, we show here that wmOPCs in vitro may resemble COPs more than gmOPCs. Indeed, COPs are hardly present Figure 6. IFNγ decreases the maximum number of intersections in gmOPCs and wmOPCs and increases the distance of the soma with the maximum number of intersections of gmOPCs. Oligodendrocyte progenitor cells (OPCs) isolated from the cortex (gmOPCs) and non-cortex (wmOPCs) of neonatal rat forebrains were left untreated or treated with 10 ng/ml TNFα, 500 U/ml IFNγ, or a combination of TNFα and IFNγ for 48 hours in the presence of PDGF-AA and FGF-2. OPCs were stained for the OPC cell surface marker A2B5 to visualize their morphology (see Fig. 4e). The number of processes that intersect with the concentric circles of Sholl analysis as a function of the distance from the soma (a,b), maximum intersections (c,e) and distance of the soma with the maximum number of branch points (d,f) are shown. Grey bars represent gmOPCs (a,c,d), white bars represent wmOPCs (b,e,f). Error bars show the standard error of the mean. Note that the maximum number of intersections is decreased in wmOPCs upon treatment with IFNγ, and in both gmOPCs and wmOPCs upon treatment with IFNγ combined with TNFα, while the distance of the soma with the maximum number of intersections is increased upon IFNγ treatment in gmOPCs. Statistical analyses were performed using a one-way ANOVA with a Tukey's post-test (*p < 0.05, **p < 0.01, ***p < 0.001) to test for differences between treatments and their respective control. Scale bar is 50 µm. in the GM sensory cortex, but abundantly present in the corpus callosum (WM) 35 . This indicates that OPCs in vivo are also more mature in the WM than in the GM, and wmOPCs may have started to exit the cell cycle and progressed to these committed OPCs. Of interest in this respect is that the mRNA level of the transcription factor Hes1, a negative regulator of myelination was higher in wmOPCs than in gmOPCs, which may keep the more mature and less proliferative wmOPCs undifferentiated in the presence of PDGF-AA and FGF-2. Indeed, it has been reported that combined exposure to these mitogens induces Hes1 expression in OPCs 27 , which may be more potent in wmOPCs. Also, wmOPCs have reduced Pdgfra mRNA levels compared to gmOPCs, which may explain why wmOPCs proliferated less in response to PDGF-AA and FGF-2 than gmOPCs.
OPC migration, proliferation and differentiation are critical for successful remyelination. While the immaturity and the slower differentiation kinetics of gmOPCs seem disadvantageous to remyelination, the opposite may be true. Adult OPCs that are activated upon demyelination return first to a more immature, neonatal-like state and this 'dedifferentiation' enhances their migratory capacities in vitro 52 . Also, upon chronic cuprizone-induced demyelination, remyelination is faster in the cerebral cortex than in the corpus callosum 11 . In contrast, a recent genetic fate mapping study demonstrate that OPC maturation upon acute cuprizone-mediated demyelination occurs slower in the cingulate cortex and hippocampus than in the corpus callosum 54 , indicating that also within GM areas OPC maturation and remyelination efficiency are heterogeneous. OPCs in the GM and WM are surrounded by different cellular and molecular environments and therefore influenced by distinct regional cues that may orchestrate OPC maturity. In addition, the origin of OPCs may signify functional differences in OPCs in GM and WM. The brain is populated by three sequential OPC waves that are generated from different regions of the forebrain ventricular zone 55 . At postnatal day 2 the third wave has populated the cortex, but not yet WM areas. This may account for intrinsic differences in neonatal gmOPCs and wmOPCs. Hmbs was used as reference gene; the reference gene Eef1a1 showed similar results (data not shown). Note that Ifngr1 expression levels are elevated in wmOPCs compared to gmOPCs. Bars represent mean relative to their respective untreated control, which was set at 1 for each independent experiment (horizontal line). Error bars show the standard error of the mean. Statistical analyses were performed using a one-sample t-test (*p < 0.05) to test for differences between treatments and their respective control and a one-way ANOVA with a Šidák post-test was used to test whether the response to TNFα, IFNγ and TNFα and IFNγ combined differed between gmOPCs and wmOPCs (not significant). Scale bar is 50 µm.
SCieNTifiC REPORtS | (2018) 8:2113 | DOI:10.1038/s41598-018-19934-6 Differences in maturation stage of gmOPCs and wmOPCs may also explain their differential responsiveness to secreted factors from normal astrocytes. In response to ACM, wmOPC migration was decreased and wmOPC differentiation was increased compared to NCM-exposed control wmOPCs, while gmOPCs were less responsive to ACM than wmOPCs. It should be noted that in the present study, ACM was derived from non-activated astrocytes, while in MS lesions astrocytes become reactive. Of interest, two distinct subsets of reactive astrocytes have been described, neurotoxic A1 and neurotrophic A2 astrocytes 56 . In contrast to the conditioned medium of normal astrocytes, neurotoxic A1 astrocytes inhibit gmOPC proliferation and differentiation 56 , while ACM of LPS-activated astrocytes also inhibit gmOPC differentiation via secreted TNFα 57 . Whether secreted factors from reactive astrocytes in the inflammatory MS lesion environment differentially affect gmOPC and wmOPC behaviour and whether this is of relevance to remyelination (failure) remains to be determined.
Upon demyelination OPCs face and respond to inflammatory mediators. The transient expression of the pro-inflammatory cytokines TNFα and IFNγ coincides with demyelination 48,58,59 , while their accumulation in MS lesions is suggested to associate with OLG cell death 24,49 . Although TNFα and IFNγ have been described to be cytotoxic to gmOPCs 41,48,60 , in the present study no increase in gmOPC and wmOPC cytotoxicity was found at defined treatment conditions. Remarkably, while TNFα was seemingly ineffective, in IFNγ-treated wmOPCs a decrease in MTT reduction and proliferation was observed, consistent with previous findings in gmOPCs 61 . In addition, upon IFNγ exposure an increase in process length was noticed in both gmOPCs and wmOPCs, while a decrease in process ramification and decelerated differentiation was observed in wmOPCs only. Strikingly, brief and transient exposure of OPCs to IFNγ is sufficient to delay wmOPC differentiation, an effect that was potentiated by TNFα. One of the few in vitro studies with wmOPCs shows that continuous exposure to IFNγ perturbs differentiation, while in the current study transient exposure of wmOPCs to IFNγ had a long-term effect on differentiation. Hence, while retaining their ability to differentiate, brief exposure to IFNγ or IFNγ and TNFα of resident OPCs in WM MS lesions may delay their differentiation, which is evidently of relevance for the development of therapeutic strategies aimed at restoring remyelination. . Note that brief exposure to IFNγ at the OPC stage delays the differentiation of wmOPCs, but not of gmOPCs, while combined treatment with TNFα and IFNγ inhibited differentiation of either OPC. Grey bars represent gmOPCs, white bars represent wmOPCs (b,c,d). Error bars show the standard error of the mean. Bars represent mean relative to their respective untreated control, which was set at 1 for each independent experiment (horizontal line). Statistical analyses were performed using a one-sample t-test (*p < 0.05) to test for differences between treatments and their respective control and a oneway ANOVA with a Šidák post-test was used to test whether the response to TNFα, IFNγ and TNFα and IFNγ combined differed between gmOPCs and wmOPCs (not significant). Scale bar is 50 µm. Taken together, neonatal gmOPCs and wmOPCs display their own distinct identity in vitro, as, among others, reflected by differences in morphology, maturity and responses to environmental (injury) signals, including pro-inflammatory cytokines. Unravelling the underlying molecular mechanisms is not only crucial for understanding OPC heterogeneity, but also for the development of therapeutic interventions, as distinct strategies may be needed to restore remyelination in GM or WM MS lesions.

Primary cell cultures. Oligodendrocyte progenitor cells (OPCs). Animal protocols were approved by the
Institutional Animal Care and Use Committee of the University of Groningen (the Netherlands). All methods were carried out in accordance with national and local experimental animal guidelines and regulations. OPCs were isolated from the neonatal cortex (Fig. 1a, referred to as gmOPCs) and neonatal non-cortical parts (WM tracts including corpus callosum, mixed GM and WM tracts, including hippocampus and thalamus, and deep GM parts, including basal ganglia, Fig. 1a, referred to as wmOPCs) of rat forebrains using a shake-off procedure as described previously 62,63 . A detailed description of the method is provided in the supplementary information. The enriched OPC fraction contained 95-97% OPCs (Olig2-positive), less than 1% microglia (IB4positive), 1-3% astrocytes (GFAP-positive) and less than 1% neurons (TuJ1-positive) for both gmOPCs and wmOPC cultures. OPCs were cultured on 13-mm poly-L-lysine (PLL, 5 µg/ml)-coated glass slides in 24-well plates unless stated otherwise. With the exception of migration and adhesion assays, cells were plated at a density of 30.000 (GM) or 40.000 (WM) cells per well in defined Sato medium 62 (see supplementary information). OPCs were synchronized to the bipolar early OPC stage by addition of 10 ng/ml platelet-derived growth factor-AA (PDGF-AA; Peprotech, cat. no. 100-13A) and 10 ng/ml human fibroblast growth factor-2 (FGF-2; Peprotech, cat. no. 100-18B) 1 hour after plating. Where indicated, 1 hour after plating cells were exposed for 48 hours to cytokines TNFα (10 ng/ml) and/or IFNγ (500 U/ml). After 2 days OPCs were allowed to differentiate in Sato medium supplemented with 0.5% fetal bovine serum (FBS) for 3 (immature oligodendrocytes (OLGs)) or 6 days (mature OLGs).
Astrocytes. Remaining astrocytes of the mixed glia cell culture flasks (see supplementary information) were passaged once by trypsinization and transferred to 162 cm 2 flasks and cultured in astrocyte medium (100 U penicillin and streptomycin, 4 mM L-glutamine, 10% heat-inactivated FBS (Bodinco, cat. no. 4005-BDC-0814) in DMEM). The enriched astrocyte fraction yielded a highly pure >97% astrocyte population. Upon reaching confluency, cells were trypsinised and plated in 6-well plates at 1*10 6 cells per well in astrocyte medium. After one day cells were washed with PBS and cultured for 24 hours in Sato medium. ACM was collected, filtered using a 0.45 µM filter and stored at −20 °C until further use. OPCs were cultured in ACM with a 1:1 ratio with Sato supplemented with 0.5% FCS where indicated. Morphology. Morphological analysis was performed as described by Langhammer and colleagues 26 . Briefly, A2B5-immunolabeled OPCs were photographed using a conventional immunofluorescence microscope (Leica DMI 6000 B), with 20× objective and photos were converted to 8-bit TIFF files. TIFF files were loaded in FIJI 67 and cellular processes were traced using the NeuronJ plugin 68 . Traces were converted to SWC files by the Bonfire-program written for MATLAB 26 . SWC files were adapted in NeuronStudio 69 after which Sholl-analysis and measurements of other morphological endpoints were performed by drawing concentric circles around the cell body with an incrementing radius of 6 µm. In each independent experiment 24-26 cells were analysed per condition. Mean values of each independent experiment were taken and plotted.
Survival assay. OPCs were plated in PLL-coated 24-well plates (Nunc; Thermo Fisher Scientific, cat. no. 144530) in triplicate. OPC survival upon cytokine exposure was assessed by 3-(4,5-Dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT; Sigma-Aldrich, cat. no. M2128)-reduction and lactate dehydrogenase (LDH; Roche, cat. no. 11644793001) assays. For the MTT-reduction assay, 500 μg/ml MTT was added to each well and left to incubate for 4 hours at 37 °C. Cells were resuspended in dimethyl sulfoxide and absorption was measured at 570 nm. LDH assays were performed according to manufacturer's instructions on medium of cells analysed in the MTT-reduction assay and related to medium of lysed untreated cells.

Migration assay.
OPCs were plated at a density of 1 × 10 5 on a PLL-coated porous membrane of a transwell insert with a pore size of 8 μm (Falcon, cat. no. 734-0053). A chemoattractive gradient was created by the addition of PDGF-AA (10 ng/ml) under the transwell insert. OPCs were allowed to migrate for 4 hours after which cells were fixed for 20 minutes in ice cold 5% acetic acid in ethanol. Cells were washed once with PBS. Cells on top of the transwell membrane were removed using a cotton swab, and nuclei of migrated cells were stained with 1 µg/ml DAPI. After washing thrice with PBS, the membranes were cut from the transwell insert and mounted in Dako mounting medium under a glass coverslip. Fluorescent images of the whole membrane were taken with the TissueFAXS fluorescent microscope. The average number of migrated cells per mm 2 was calculated using Tissuequest 4.0 software. Total number of migrated cells was calculated from the total surface of the membrane and related to total plated cells after correction for non-adhering cells via the adhesion assay.
Adhesion assay. Cells were plated at density of 1 × 10 5 in PLL-coated wells in triplicate on a 96-well plate (Nunc; Thermo Fisher Scientific, cat. no. 167008) in 50 μl Sato medium. After one hour, PDGF-AA was added. After 4 hours, cells were fixed with ice cold methanol for 10 minutes. Cells were washed with PBS after which 0.2% crystal violet solution in ethanol was added for 10 minutes. Wells were then washed thrice with water and cells were dissolved in 1% sodium dodecyl sulphate. Absorption was measured at 570 nm after 30 minutes. Adhesion of gmOPCs was set to 1 for further analysis.
Statistical analysis. Data are expressed as mean ± standard error of the mean (SEM) for at least three independent experiments. When absolute values between two groups were compared (i.e., gmOPCs vs wmOPCs) statistical significance was assessed using a paired two-sided t-test and when more than two groups were compared a one-way ANOVA followed by a Tukey's post-test was used. Statistical analysis was performed with a one-sample t-test when relative values of groups were compared by setting the untreated control values at 1 at each independent experiment. When relative values of two conditions were compared between gmOPCs and wmOPCs an unpaired two-sided t-test was used. When relative values of multiple treatment conditions were compared between gmOPCs and wmOPCs a one-way ANOVA with a Šidák post-test was used. Statistics were performed using GraphPad Prism 6.0. In all cases p-values of <0.05, <0.01, and <0.001 were considered significant and indicated with *, **, *** respectively.
Data availability. All data generated during and/or analysed during the current study are available from the corresponding author on reasonable request.