(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



A molecularly defined subpopulation of oligodendrocyte precursor cells controls the generation of myelinating oligodendrocytes during postnatal development [1]

['Shayan Moghimyfiroozabad', 'Center For Interdisciplinary Research In Biology', 'Cirb', 'College De France', 'Cnrs', 'Inserm', 'Université Psl', 'Paris', 'Maela A. Paul', 'Lea Bellenger']

Date: 2024-07 

Oligodendrocyte precursor cells (OPCs) are a class of glial cells that uniformly tiles the entire central nervous system (CNS). They play several key functions across the brain including the generation of oligodendrocytes and the control of myelination. Whether the functional diversity of OPCs is the result of genetically defined subpopulations or of their regulation by external factors has not been definitely established. We discovered that a subpopulation of OPCs found across the brain is defined by the expression of C1ql1, a gene previously described for its synaptic function in neurons. This subpopulation starts to appear during the first postnatal week in the mouse cortex. Ablation of C1ql1-expressing OPCs in the mouse leads to a massive lack of oligodendrocytes and myelination in many brain regions. This deficit cannot be rescued, even though some OPCs escape Sox10-driven ablation and end up partially compensating the OPC loss in the adult. Therefore, C1ql1 is a molecular marker of a functionally non-redundant subpopulation of OPCs, which controls the generation of myelinating oligodendrocytes.

A fundamental question is whether the functional diversity of OPCs arises from molecularly defined subpopulations with distinct functions or whether it is a result of their interaction with the environment [ 1 , 17 ]. In this study, we show that expression of C1ql1, a gene known for its role in neuronal synapse formation [ 22 , 23 ], marks a subpopulation of OPCs throughout the mouse brain starting around birth. Using genetic ablation, we show that the elimination of this subpopulation leads to lack of OLs in the dorsal forebrain and a massive reduction of myelination. While some OPCs escape Sox10-driven ablation and end up compensating the lack of OPCs with time, they cannot rescue OL and myelination deficits. Thus, we identify a molecular signature for a subpopulation of OPCs that generates myelinating OLs during postnatal development of the mouse forebrain.

OPCs are generated in the CNS through sequential waves and in a ventro-dorsal manner [ 7 ]. In the forebrain, 3 waves with different spatial origins and genetic lineages generate OPCs [ 12 ]. The first 2 waves are originally ventral and embryonic. The third wave is generated dorsally after birth [ 12 ]. The elimination of each wave of OPCs is rapidly compensated by the other waves without any significant consequences on OPC dynamics or myelination of the developing brain [ 12 ]. Even the ablation of all 3 waves in the forebrain is compensated by OPCs coming from the diencephalon within 2 weeks [ 7 , 13 ]. Thus, despite their different origins, various waves of OPCs seem to be functionally redundant. However, many data also suggest functional heterogeneity in OPCs [ 14 ]. Migration of OPCs towards the optic nerve can be Nrg1-ErbB4-signaling dependent or independent depending on whether OPCs are generated from the first or second wave [ 15 ]. Another example is the Shh-dependent modulation of the generation of OPCs during the first wave, while the other waves are Shh-independent [ 16 ]. Independent of their origin, OPCs have different morphologies based on their location [ 1 , 17 ]. OPCs in the white matter have an elongated soma packed tightly between the myelinated axons, with processes extending parallel to the direction of the fibers. OPCs in the gray matter have a star-shaped morphology with several long and thin branches extending in all directions [ 1 ]. OPCs can also have different dynamics in terms of proliferation, migration, and differentiation rate [ 8 ]. OPCs in the corpus callosum differentiate into oligodendrocytes faster than OPCs in the cortex [ 8 ]. OPCs in ventral and dorsal forebrain respond differently to DNA damage by promoting senescence or apoptosis, respectively [ 18 ]. A single-cell RNAseq analysis of OPCs isolated from the mouse forebrain at different stages during development (E13.5, P7, and P21) showed that OPCs originating from several regions of the forebrain become transcriptionally equivalent at P7 [ 19 ]. However other studies have shown that OPCs are heterogeneous based on their transcriptomes, their surfaceomes (e.g., receptors and ion channels) or electrophysiological properties [ 17 , 20 , 21 ]. Spitzer and colleagues showed that OPCs start to express voltage-gated ion channels and glutamate receptors differentially throughout development in concordance with their functional states [ 21 ]. All these studies demonstrate that OPCs are a diverse population and their diversity changes based on the brain region and age.

Oligodendrocyte precursor cells (OPCs) are a type of glial cells that homogeneously populate the central nervous system (CNS) starting during postnatal development and throughout adulthood [ 1 ]. Several functions have been attributed to OPCs [ 2 ]. They have been documented to control vascularization and angiogenesis [ 3 ], engulfment of axons and refinement of neural circuits [ 4 , 5 ], and antigen presenting [ 6 ]. However, a major, and their initially described, role is to differentiate into oligodendrocytes (OLs) [ 7 – 9 ]. In rodents, the generation of OLs from OPCs starts at birth and continues during adulthood [ 10 , 11 ]. OPC differentiation peaks during the second postnatal week and then decreases gradually [ 10 ]. A portion of OPCs continues to proliferate to maintain the density of their population constant throughout the mature brain [ 1 ]. OPCs are thus essential for proper development and function of brain circuits.

Results

A subpopulation of OPCs is characterized by C1ql1 expression Breeding of the C1ql1Cre knockin mouse model [24] with reporter mouse lines can be used to determine the history of expression of C1ql1 during brain development. Using the Cre-dependent R26Cas9-GFP reporter mouse line ([24,25] and Fig 1A), we noticed that, in addition to the expected expression in certain neuronal populations [24], GFP was expressed by a category of cells present throughout the brain at postnatal day 30 (P30) (Fig 1B). These cells did not appear to be neurons, an observation that was confirmed using immunolabeling for NeuN (S1 Fig). While cerebellar granule cells and inferior olivary neurons were both GFP and NeuN positive as expected (S1 Fig), the other GFP-labeled cells were NeuN negative. Because the morphology of these cells was suggestive of glia, we tested markers for various glial populations. The GFP-positive cells were negative for GFAP showing that they were not astrocytes (S1 Fig). OPCs are characterized by the expression of neural/glial antigen 2 (NG2) and platelet-derived growth factor receptor α (PDGFRa). These markers immediately disappear upon differentiation into OLs [1,9]. Using NG2 immunolabeling, we determined that the non-neuronal cells labeled using the C1ql1Cre knockin mouse line were a subpopulation of OPCs: some, but not all, NG2-labeled cells were also GFP-positive (Fig 1B). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. A subpopulation of oligodendrocyte precursor cells is characterized by C1ql1 expression. (A) Diagram of the breeding scheme used to determine the cell types expressing C1ql1 in the mouse brain. The C1ql1Cre mouse line was crossed with the Cre-dependent R26Cas9-GFP reporter line. (B) GFP-expressing cells and OPCs were visualized using co-immunolabeling for GFP (green) and NG2 (magenta) respectively, in parasagittal brain sections from C1ql1Cre; R26Cas9-GFP animals at postnatal day 30 (P30). In the cerebellum and the brainstem, GFP is strongly expressed by granule cells and inferior olivary neurons, respectively. Scale bars = 50 μm. Higher magnifications of the regions delineated by the white squares show the presence of GFPneg (white arrowhead) and GFPpos (yellow arrowhead) NG2 cells. Scale bars = 20 μm. (C) Duplex smFISH of C1ql1 and Cspg4 (coding for NG2) mRNAs was performed in parasagittal sections of the somatosensory cortex, cerebellum, and brainstem from P30 wild-type animals. Both C1ql1neg (white arrowhead) and C1ql1pos (yellow arrowhead) Cspg4pos cells are detected. Scale bars = 50 μm. Bottom: Quantification of the percentage of C1ql1pos Cspg4pos cells was performed in the somatosensory cortex (Crtx), cerebellum (Cb), and brainstem (BS). Data are represented as mean ± SEM. n = 4 animals. (D) New analysis of the single-cell transcriptomics raw data from Marques and colleagues [19] reveals the different cell clusters of the oligolineage (left panel) and confirms the expression of C1ql1 (right panel) in a subset of OPCs and COP cells. COP, committed oligodendrocyte precursors; eVLMC, embryonic vascular and leptomeningeal cells; NFOL, newly formed oligodendrocyte; NP, neural progenitor; OPC, oligodendrocyte precursor cell; pnVLMC, postnatal vascular and leptomeningeal cells; smFISH, single-molecule fluorescence in situ hybridization. The individual numerical values underlying this figure can be found in S1 Data. https://doi.org/10.1371/journal.pbio.3002655.g001 Our genetic approach suggests that a subpopulation of OPCs, and/or their progenitors, express C1ql1. We performed duplex single-molecule fluorescence in situ hybridization (smFISH) for C1ql1 and Cspg4 (coding the marker NG2) on brain sections from P30 mice (Fig 1C). C1ql1 puncta were observed together with Cspg4 puncta in a subset of cells throughout the brain. Quantification showed that about 57% (1,476/2,542 cells), 50% (1,122/2,232 cells), and 61% (2,099/3,454 cells) of Cspg4 expressing cells in the cortex, cerebellum, and brainstem, respectively, express C1ql1 (Fig 1C). Single-cell transcriptomics of the oligodendrocyte lineage at embryonic (E13.5) and postnatal stages (P7, juvenile, and adult) has previously shown that, besides a subpopulation of cycling OPCs, OPCs form a rather homogenous cell population at the transcriptome level at P7 [19]. Using the raw expression data from these studies [19], we performed our own bioinformatic analysis to determine the pattern of C1ql1 expression across the different cell clusters constituting the oligodendrocyte lineage. We used a batch correction to eliminate the effects due to the different experimental batches (E13.5 and P7 versus juvenile and adult, cf. S2 Fig), followed by uniform manifold approximation and projection (UMAP) reduction and clustering on shared K-nearest neighbor (SNN) graph. This pipeline led to the identification of 10 clusters including 3 OPC subpopulations (Fig 1D). Two of these subpopulations were the same as the ones found by Marques and colleagues [19] (OPC1 and the cycling OPC subpopulations), while a small third one (OPC2) was defined by the enriched expression of genes associated with the GO terms “mRNA processing” and “mRNA splicing,” suggesting a specific metabolism of RNA. C1ql1-expressing cells were found in all 3 OPC clusters as well as in a subset of committed OPCs (COPs) that are differentiating into oligodendrocytes (Fig 1D). C1ql1 expression was almost undetected in newly formed oligodendrocytes (NFOL). Thus, genetic and transcriptomic data altogether demonstrate that a subpopulation of OPCs expresses C1ql1 in the juvenile brain and raise the question as to whether this subpopulation of OPCs has a specific role in the brain.

Ablation of C1ql1-expressing OPCs is not compensated during postnatal development In the adult mouse brain, OPCs maintain their homeostasis by local proliferation [26]. Ablation, differentiation, or apoptosis of OPCs leads to proliferation and migration of the adjacent cells to replace the missing ones [26]. OPCs that are generated in different waves show functional redundancy, as ablation of one wave using Sox10-driven expression of diphtheria toxin (DTA) is compensated by the other 2 waves in the cortex during the first 2 postnatal weeks [12]. Even ablation of all 3 waves upon their generation during embryonic development and birth induces repopulation by the OPCs from other regions of the brain (e.g., diencephalon) within 15 days [7,13], showing the strong capacity of OPCs for sensing OPC disappearance and compensating for it [12,13,26]. We used the previously established Sox10-DTA based ablation method to test whether C1ql1-expressing OPCs (C1ql1pos OPCs) are functionally redundant with other OPCs. For this, the C1ql1Cre mouse line was crossed with the Sox10DTA line expressing DTA in a Cre-dependent manner [12] (Fig 2A). The Sox10 promoter is specifically active in all the oligodendroglial lineage and drives GFP expression in more than 98% of the oligolineage cells in the Sox10DTA line [12]. Accordingly, in the absence of Cre (C1ql1wt; Sox10DTA mice), the vast majority of the oligolineage cells, OPCs, COPs, and OLs, are GFP positive (Fig 2B). The expression of the Cre recombinase using the C1ql1Cre locus leads to loss of GFP, induction of DTA expression, and death of the Cre-expressing cells in the oligolineage (Fig 2A). C1ql1 is not expressed in progenitors (Fig 1D), and thus, expression of Cre does not affect the generation of C1ql1-expressing neurons. The C1ql1Cre; Sox10DTA mice only rarely survived beyond 3 to 4 postnatal weeks. Because C1ql1 and Sox10 are co-expressed in various cell types of the enteric nervous system (http://mousebrain.org/adolescent/genesearch.html), this lethality might be due to its dysfunction, as has been described in other mutants with enteric system dysfunction [27]. All the animals survived until P16 without any major visually detectable behavior or motor deficit. Morphological analysis showed a massive reduction in the number of GFP-expressing cells all over the brain, and in particular in the forebrain, after ablation of C1ql1-expressing OPCs (C1ql1Cre; Sox10DTA) compared to controls (C1ql1wt; Sox10DTA, Fig 2B–2D). Since the development of oligolineage cells has been well studied in the forebrain, particularly in the corpus callosum and the cortex [12], we selected the central and lateral parts of the corpus callosum (CC-center and CC-lateral), as well as the retrosplenial area of the cortex for detailed analysis (Fig 2C and 2D). The number of Sox10-GFP positive (GFPpos) cells is reduced by 66%, 85%, and 46% in these 3 regions, respectively, upon C1ql1Cre-induced ablation (Fig 2C; mean ± SEM in the control and ablation conditions: CC-center = 41.2 ± 2.1 versus 13.9 ± 1.2 × 104 cells/mm3; CC-lateral = 39.1 ± 3.0 versus 5.7 ± 0.6 × 104 cells/mm3; cortex = 10.0 ± 1.5 versus 5.4 ± 0.5 × 104 cells/mm3). Since GFP labels nearly the entire oligolineage cells in the Sox10DTA mouse model [12], the OPC marker NG2 [1,9] was used to quantify the OPC subpopulation (NG2pos GFPpos cells). A 57%, 80%, and 32% reduction in the density of NG2pos GFPpos cells was found in the center and lateral parts of the corpus callosum, and cortex, respectively (CC-center: 25.0 ± 4.1 versus 10.7 ± 1.0; CC-lateral: 24.3 ± 3.2 versus 4.8 ± 0.6 × 104 cells/mm3; cortex: 7.4 ± 1.6 versus 5.0 ± 0.5 × 104 cells/mm3). In comparison with NG2 staining, immunolabeling for PDGFRa, another marker of OPCs [1,9], is more concentrated on the somata of OPCs and allows easier quantification of individual OPCs (Fig 2C and 2D). These quantifications also showed a reduction in the density of PDGFRapos GFPpos cells in all 3 regions at P16 (CC-central: 74% decrease, mean ± SEM = 20.1 ± 1.2 versus 5.3 ± 0.7 × 104 cells/mm3; CC-lateral: 79%; 26.5 ± 3.3 versus 5.7 ± 0.4 × 104 cells/mm3; cortex: 29%; 7.9 ± 0.8 versus 5.6 ± 0.4 × 104 cells/mm3). Altogether, using the 2 most common markers of OPCs, NG2 and PDGFRa, these results confirm that DTA expression driven by the C1ql1Cre knockin line leads to a massive disappearance of OPCs in several regions of the forebrain. They also suggested that, in contrast to what has been reported when ablating OPCs from different waves [12,13], remaining OPCs cannot compensate for the loss of C1ql1-expressing OPCs during postnatal development. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Ablation of C1ql1-expressing OPCs is not compensated during postnatal development. (A) Diagram of the breeding scheme used to ablate C1ql1-expressing oligolineage cells in the mouse brain via Sox10-driven expression of diphtheria toxin (DTA). The C1ql1Cre mouse line was crossed with the Cre-dependent Sox10DTA mice. (B) Parasagittal and coronal sections from P16 brains of control (C1ql1wt; Sox10DTA) and C1ql1pos OPC-ablated (C1ql1Cre; Sox10DTA) mice. Direct GFP fluorescence and Hoechst staining are shown. Scale bars = 1 mm. Numbers indicate the regions of the corpus callosum (1: CC-center, 2: CC-lateral) and cortex 3: used for quantifications. (C) OPCs were immunolabeled for NG2 (magenta) in coronal sections from P16 brains of the control and C1ql1-ablated conditions. Both NG2pos (yellow arrowhead) and NG2neg (white arrowhead) GFPpos cells are detected in the center and lateral part of corpus callosum (CC) as well as the cortex. Scale bars = 100 μm. The mean densities of both NG2pos GFPpos (OPCs) and NG2neg GFPpos oligolineage cells are reduced in C1ql1Cre; Sox10DTA brains. Data are represented as mean ± SEM. C1ql1wt; Sox10DTA: n = 4 animals. C1ql1Cre; Sox10DTA: n = 7 animals. Unpaired t test with Welch’s corrections. ns = not significant. (D) OPCs were immunolabeled for PDGFRa (magenta) in coronal sections from P16 brains of the control and C1ql1-ablated mice. Both PDGFRapos (yellow arrowhead) and PDGFRaneg (white arrowhead) GFPpos cells are detected. Scale bars = 100 μm. Quantification confirms the reduced density of both PDGFRapos GFPpos (OPCs) and PDGFRaneg GFPpos oligolineage cells upon ablation of C1ql1-expressing Sox10pos cells. Data are represented as mean ± SEM. C1ql1wt; Sox10DTA: n = 6 animals. C1ql1Cre; Sox10DTA: n = 10 animals. Unpaired t test with Welch’s corrections. The individual numerical values underlying this figure can be found in S1 Data. OPC, oligodendrocyte precursor cell; PDGFRa, platelet-derived growth factor receptor α. https://doi.org/10.1371/journal.pbio.3002655.g002

C1ql1 expression in OPCs follows region-specific time-courses during postnatal development Lack of compensation of OPC loss by P16 in C1ql1Cre; Sox10DTA brains could be due to a reduced amount of time since ablation compared to previously published results [12,13,28]. Furthermore, the magnitude of OPC loss at P16 is bigger in the corpus callosum than in the cortex in the C1ql1Cre; Sox10DTA mice, suggesting regional differences. We analyzed the timing of expression of C1ql1 in OPCs during postnatal brain development. Analysis of previously published data of single-cell transcriptomics of oligolineage cells [19] shows a negligible expression of C1ql1 at E13.5, while a considerable number of OPCs express C1ql1 already by P7 and in juvenile animals (S3A Fig). This suggests that C1ql1 expression is induced during the first postnatal week in a subpopulation of OPCs in the forebrain. We performed duplex smFISH for C1ql1 and Cspg4 on brain sections from wild-type mouse at P7 and P15 (Fig 3A). Similar to our observations at P30 (Fig 1C) and in accordance with the single-cell transcriptomic data (S3A Fig), C1ql1 labeling was detected in a subgroup of Cspg4pos cells already by P7 (Fig 3A). Quantification demonstrated that 72% (63,626/88,314 cells) and 68% (77,714/113,512 cells) of Cspg4pos cells in the cortex express C1ql1 at P7 and P15, respectively, a percentage higher by about 10% than the percentage quantified at P30 (Fig 1C). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. Lack of oligodendrocytes after DTA-mediated ablation of C1ql1-expressing OPCs. (A) Duplex smFISH of C1ql1 and Cspg4 mRNAs was performed in coronal sections from the cortex of P7 and P15 wild-type mice. Both C1ql1neg (white arrowhead) and C1ql1pos (yellow arrowhead) Cspg4pos cells are observed. Scale bars = 50 μm. Quantification of the percentage of C1ql1-expressing Cspg4pos cells was performed in the cortex of P7 and P15 animals. Data are represented as mean ± SEM. P7: n = 3 animals. P15: n = 4 animals. Scale bars = 50 μm. (B) GO analysis shows the biological functions of the genes enriched in C1ql1-expressing OPCs compared to other OPCs (reanalysis of raw data from Marques and colleagues [19]). (C) Coronal sections from C1ql1wt; Sox10DTA and C1ql1Cre; Sox10DTA brains at P16 were immunolabeled for CC1, a marker of oligodendrocytes (magenta). Scale bars = 500 μm. Higher magnification of the coronal sections in the center and lateral parts of CC and cortex shows the dramatic reduction in the density of OLs (CC1pos GFPpos cells, yellow arrowhead) in C1ql1Cre; Sox10DTA mice at P16. Scale bars = 100 μm. Quantification shows a massive decrease in OL density in the C1ql1pos OPC-ablated brains. Data are represented as mean ± SEM. C1ql1wt; Sox10DTA: n = 5 animals. C1ql1Cre; Sox10DTA: n = 4 animals. Unpaired t test with Welch’s corrections. ns = not significant. (D) Coronal sections from C1ql1wt; Sox10DTA and C1ql1Cre; Sox10DTA cortex from adult mice were immunolabeled for CC1. Scale bars = 500 μm. Higher magnification shows rare CC1pos GFPpos Sox10 lineage-derived OLs in the corpus callosum of C1ql1Cre; Sox10DTA adult brains, and repopulation by GFPneg OLs (GFPneg CC1pos cells, white arrowhead) only in the center of the corpus callosum. In the cortex, cells that do not seem to be OLs are CC1 positive (probably astrocytes as seen in [34]). Scale bars = 100 μm. The individual numerical values underlying this figure can be found in S1 Data. GO, Gene Ontology; OPC, oligodendrocyte precursor cell; smFISH, single-molecule fluorescence in situ hybridization. https://doi.org/10.1371/journal.pbio.3002655.g003 How early does C1ql1 expression appear in OPCs? smFISH showed the presence of C1ql1 mRNA in a subgroup of Cspg4pos cells already by P1 in the cortex (S3B Fig). However, due in particular to the high density of cells in the forebrain at that age, accurate quantification was not possible. To assess the regional differences in the timing of C1ql1 expression, we thus used a genetic approach and crossed the C1ql1Cre line with the Cre-dependent R26R-EYFP reporter line [29]. Co-immunolabeling for YFP, PDGFRa, and NG2 was performed on coronal forebrain slices of P1-P2 C1ql1Cre; R26R-EYFP/WT mice (S3C Fig). About 40% of OPCs expressed YFP in both regions of the corpus callosum by P1-P2 while only 10% in the cortex (S3C Fig; mean ± SEM = 42.4 ± 8.3% versus 35.6 ± 4.1% versus 10.5 ± 0.6% in the central part, lateral part of corpus callosum, and the cortex, respectively). Altogether, our results show that the timing of appearance of C1ql1 expression in OPCs is region-dependent, with earlier expression in the corpus callosum than in the cortex. In the cortex, this expression appears around birth and increases during the 2 first postnatal weeks of development.

Loss of C1ql1-expressing OPCs is compensated by cells escaping Sox10-driven ablation Does the different time-course of C1ql1 expression in OPCs across forebrain regions translate into a different time-course of OPC loss in C1ql1Cre; Sox10DTA mice? In the corpus callosum, the percentage of OPCs expressing C1ql1 at P16 (roughly 70%) is in the range of the OPC loss observed in C1ql1Cre; Sox10DTA mice (70% to 80%, Fig 2). In the cortex, the density of OPC is reduced by only 30% at the same age. Quantification did not reveal changes in the density of oligolineage (GFPpos) cells in the brains of C1ql1Cre; Sox10DTA mice at P0 (S4A Fig; 19.4 ± 1.1 versus 23.7 ± 4.5 in CC-center, 32.1 ± 1.9 versus 25.7 ± 3.0 in CC-lateral, and 18.5 ± 1.0 versus 17.7 ± 1.2 × 104 cells/mm3 in the cortex). At this age, the density of OPCs (PDGFRapos GFPpos cells) was reduced in the lateral part of the corpus callosum (S4A Fig; 34% decrease; mean ± SEM = 12.9 ± 0.5 versus 8.4 ± 1.4 × 104 cells/mm3), but not in the center part or in the cortex. At P7, the density of oligolineage cells is already decreased by 40% to 50% in the corpus callosum, while no change is detected in the cortex (S4B Fig, CC-Center: mean ± SEM = 32.7 ± 1.9 in the control versus 16.3 ± 1.5 × 104 cells/mm3; CC-lateral: 45.9 ± 2.5 versus 26.6 ± 2.5 × 104 cells/mm3; cortex: 13.3 ± 1.0 versus 15.9 ± 3.2 × 104 cells/mm3). The density of OPCs (PDGFRapos GFPpos cells) was found to be decreased by about 55% in the center and lateral corpus callosum, while no change was detected in the cortex (S4B Fig; mean ± SEM: CC-center = 17.1 ± 1.2 versus 7.6 ± 2.5 × 104 cells/mm3; CC-lateral = 14.9 ± 1.3 versus 6.9 ± 2.1 × 104 cells/mm3; cortex = 7.1 ± 0.4 versus 6.9 ± 2.5 × 104 cells/mm3). These results show that the time-course of OPC loss in the 3 analyzed regions correlates with the time-course of appearance of C1ql1 in OPCs, with the cortex being affected later than the corpus callosum. In the cortex, OPC loss is not yet observed at the end of the first postnatal week, despite the presence of about 10% C1ql1pos OPCs already at P1-P2 and the steep increase of this percentage during the first postnatal week. The slower time-course of OPC loss in the cortex could be due to increased proliferation of remaining OPCs. To assess this, we used immunolabeling for Ki67, a marker for proliferative cells [30] (S5 Fig). No significant change in the percentage of proliferative oligolineage cells (Ki67pos GFPpos cells) was detected in the cortex at neither P0 nor P7 (mean ± SEM = 22.3 (2,308/10,692 cells) ± 2.5% versus 26.3 (2,799/10,643 cells) ± 1.8% at P0 and 5.3 (224/4,673 cells) ± 1.7 versus 7.7 (260/4,312 cells) ± 2.7% at P7). Another factor that could slow down OPC loss in this region during the first postnatal week could be the migration of newly formed OPCs, which ends only by P10 in the dorsal forebrain [12]. Regardless of these regional differences, the ablation of OPCs in C1ql1Cre; Sox10DTA mice does not seem to be compensated by the remaining OPCs during postnatal development, since a significant decrease in OPCs is detected by P16 in all 3 regions. Could compensation occur with longer recovery time? Rarely, some C1ql1Cre; Sox10DTA mice survived into adulthood (>8 weeks). Morphological analysis showed a global massive decrease in the number of GFP-expressing cells in the forebrain of these mice compared to C1ql1wt; Sox10DTA controls, similarly to our observations at P16 (S6A Fig). Immunolabeling for NG2 and PDGFRa showed only rare Sox10-GFPpos PDGFRapos NG2pos cells and the appearance of GFPneg PDGFRapos NG2pos cells in the cortex and the central and lateral parts of the corpus callosum (S6B Fig). This means that a compensation of OPC loss can occur after many weeks, but from cells in which the promoter of Sox10 has low or no activity (less than 2% of the cells [12,28]) and thus escape Sox10DTA-mediated ablation in the presence of Cre recombinase. These cells could be derived locally or from neural progenitor cells in the ventricular-subventricular zone, as observed several weeks after ablation by pharmacological and genetic tools of all OPCs in the adult mouse brain [28].

Ablation of C1ql1-expressing OPCs leads to lack of oligodendrocytes C1ql1-expressing cells are present in all OPC clusters defined by single-cell transcriptomics (Fig 1D). This implies that instead of defining a genetically distinct subgroup, C1ql1 is a marker of a specific transitional state and/or a specific function within the OPC subpopulation. We performed overrepresentation analysis of the Gene Ontology (GO) terms associated with C1ql1pos OPCs compared to other OPCs in single-cell transcriptomic data (our reanalysis of data from [19], Figs 1D, 3B, S7A and S7B). “Gliogenesis” and “glial cell differentiation” were the 2 most significantly overrepresented terms (gene ratio = 0.19 and 0.15, respectively) in C1ql1pos OPCs (Figs 3B and S7A), suggesting that C1ql1pos OPCs might already be committed to differentiation. Indeed, C1ql1 expression is also detected in some committed oligodendrocyte precursors (COPs) but disappears in NFOLs (Fig 1D). Other markers have been previously found to label a specific transition stage of OPC differentiation in OLs [31–33]. The long noncoding RNA Pcdh17it labels newly born immature OLs (as confirmed by our analysis, Figs 1D and S7B), while Gpr17 and Bcas1 are markers of COPs [32,33] (Figs 1D and S7B). C1ql1pos cells do not overlap with cells expressing Gpr17, Bcas1, and Pcdh17it markers to a significant extent, while many C1ql1pos cells coexpress Cspg4 (S7B Fig). Thus, C1ql1 expression is not another marker of COPs and NFOLs, but rather marks a specific state in a subpopulation of OPCs that is ready for differentiation into oligodendrocytes. Indeed, quantification of the number of GFPpos cells after ablation of C1ql1-expressing OPCs showed a massive decrease not only in NG2pos GFPpos cells, but also in NG2neg GFPpos cells (Fig 2). Because GFP expression is driven by the Sox10 promoter, this suggested that DTA ablation in C1ql1pos OPCs results in the reduction not only in OPC numbers but also in COPs and oligodendrocytes. Indeed, at P16 NG2neg GFPpos cells are completely absent in the cortex and corpus callosum. Using immunolabeling of P16 forebrain coronal slices for CC1, a marker of oligodendrocytes, we confirmed the massive reduction in OL numbers after ablation of C1ql1-expressing OPCs, with an almost complete absence of CC1pos cells in the cortex (Fig 3C). Quantification demonstrated a 77%, 98%, and 83% reduction in the number of OLs (CC1pos GFPpos cells) in the center (mean ± SEM = 18.9 ± 2.3 versus 4.2 ± 1.9 × 104 cells/mm3) and lateral (mean ± SEM = 15.2 ± 1.7 versus 0.2 ± 0.0 × 104 cells/mm3) corpus callosum, and in the cortex (mean ± SEM = 1.2 ± 0.3 versus 0.2 ± 0.1 × 104 cells/mm3), respectively. In the adult, several weeks after ablation of C1ql1-expressing OPCs, only very few GFPpos CC1pos cells were visible in the corpus callosum and cortex of C1ql1Cre; Sox10DTA forebrain (Fig 3D). GFPneg CC1pos OLs were present only in the center of the corpus callosum. Of note, in this region, CC1 cellular expression was also up-regulated compared to the level observed in control animals. In the cortex, cells that do not seem to be OLs are CC1 positive (probably astrocytes as seen in [34]). Thus, despite the repopulation by GFPneg OPCs in the adult C1ql1Cre; Sox10DTA brain (S6B Fig), there is no generation of OLs other than in the center of the corpus callosum. These results show that C1ql1-expressing OPCs are essential for the generation and/or survival of OLs in the brain.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002655

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/