TLR2 and TLR4 Expression Maintain the Neural Progenitor Cell Population in the Postnatal Mouse Spinal Cord
TLR2 and TLR4 displayed similar expression levels in mouse neonatal spinal cord extracts at the mRNA (Fig. 1A) and protein levels (Fig. 1B). We also found constitutive mRNA expression of other TLR family members - while TLR3 showed the highest expression levels, TLRs 6, 8, and 9 displayed lower expression levels than the other tested TLRs (Supplementary Fig. 1). The immune-histochemical analysis of spinal cords provided evidence that most TLR2-expressing cells (orange) co-expressed TLR4 (green), as shown in the representative images (Fig. 1C; white square contains higher magnification of the indicated area shown for each staining).
To further investigate the role of TLR2 and TLR4 in neonatal mouse spinal cord-resident NPCs, we explored the expression of Sox2, the earliest transcription factors expressed in neural stem and progenitor cells [30], playing a key role in specifying early neural lineages and brain development [31]. In the early postnatal WT mouse spinal cord, we observed Sox2 expression within the central canal (CC) (Fig. 1D) in cells homogeneously distributed dorsally and ventrally within the grey matter at the mantle area (Fig. 1D; MT) and within the lining along the cord perimeter in the white matter (PM) (Fig. 1D). Interestingly, the deletion of TLR2 or TLR4 significantly diminished the total number of Sox2-positive cells (Fig. 1E). When we closely inspected Sox2-expressing cells, we distinguished two different expression patterns – nuclear Sox2 expression, corresponding to dividing progenitors NPCs [32] (Fig. 1D, inset *), previously reported as potential oligodendrocyte progenitor cells (OPC) and mature astrocytes [33], and cytoplasmic Sox2 expression, corresponding to migrating and non-dividing neuroblasts [34], co-expressing NeuN (Fig. 1D, inset #). Our analysis revealed that TLR2 or TLR4 loss only affected the number of cells with nuclear Sox2 expression, assigned as the NPCs (Fig. 1F; solid blue bar), since the total number of OPC (Fig. 1G, left graph) or astrocytes (Fig. 1G, right graph) determined by the expression of Olig2 and GFAP respectively, did not show significant differences. We observed the vast majority of cells with cytoplasmic Sox2 expression located in the grey matter i.e., neuroblasts (Fig. 1H, left panel, blue striped bar) co-existing alongside cells with nuclear Sox2 expression (Fig. 1H, left, solid blue bar); however, TLR2 or TLR4 loss failed to impact these Sox2 positive cells. Likewise, all Sox2-positive cells located to the cord perimeter/meningeal zone, [35], displayed nuclear Sox2 expression and here we found a significant impact by the depletion of TLR2 or TLR4 by promoting a significant decrease on the number of this neural precursor population (Fig. 1H, right). Overall, these findings suggest that TLR2 and TLR4 support spinal cord NPC maintenance at early postnatal stages.
An analysis of cell proliferation within the spinal cords of WT and also TLR2-/- and TLR4-/- postnatal mice via Ki67 immunostaining failed to find any significant differences (Fig. 1I), suggesting that the decreased number of Sox2-expressing proliferating cells fails to significantly affect overall proliferative activity at the postnatal stage.
We also explored whether the TLR2 or TLR4 expression contributed to the FoxJ1-expressing ependymal precursor cell population. FoxJ1, a transcription factor involved in ciliogenesis [36], is considered a marker of fully differentiated and ciliated ependymal progenitor cells that line the central canal and divide and differentiate at postnatal time points [37, 38]. However, we failed to find any significant differences in the number of FoxJ1-positive cells in WT, TLR2−/−, and TLR4−/− mouse postnatal spinal cords (Fig. 1J, green). FoxJ1 positive cells co-localized with Sox2 (Fig. 1J orange) but not with GFAP (Fig. 1J, blue), as previously noted [39].
Finally, an analysis of NeuN levels demonstrated a significant reduction in the total number of neurons in TLR2−/− but not TLR4−/− mouse neonatal spinal cords than WT mice (Fig. 1K).
Overall, these studies prompted us to undertake a more detailed in vitro analysis of in vitro-expanded NPCs isolated from the spinal cords of WT, TLR2−/−, and TLR4−/− mice to decipher the specific roles of these two TLRs in NPC self-renewal and differentiation.
TLR Expression by in vitro-Expanded NPCs Isolated from the Postnatal Mouse Spinal Cord
We next explored the relative expression of TLR2 and TLR4 by in vitro-expanded NPCs derived from mouse neonatal spinal cords (schematic representation of the NPC in vitro expansion procedure shown in Fig. 2A). We failed to find significant differences in mRNA (Fig. 2B) and protein (Fig. 2D) expression levels for TLR2 and TLR4 in WT NPCs Immunofluorescence analysis confirms that all NPCs co-express for receptors (Fig. 2C). Next, we explored MyD88/TRIF-mediated responses of TLRs in WT NPCs upon stimulation with 50 ng/ml of LPS for 30 or 60 minutes (Fig. 2E). All three tested downstream mediators of TLR2 and TLR4, induced nitric oxide synthase (iNOS), phosphorylated extracellular signal-regulated kinase (pERK), and interferon regulatory factor 1 (IRF1), displayed maximal activation (as measured by an increase in protein levels for iNOS and IRF1 and phosphorylation of ERK) 30 minutes after stimulation, with a decrease 30 minutes later (Fig. 2E; representative Western blots shown on the left) as previously described in other cell types [40, 41]. These data provided evidence for the responsive and functional nature of TLR2 and TLR4 to LPS by in vitro-expanded WT NPCs derived from mouse neonatal spinal cords.
We also evaluated the expression of the TLR family members in NPCs isolated from the spinal cords of TLR2-/- and TLR4-/- neonatal mice. Our findings confirmed the lack of TLR2 and TLR4 expression in TLR2-/- and TLR4-/- mice, respectively, and additionally demonstrated that TLR2 loss significantly reduced the expression level of TLR1, TLR6, TLR8, and TLR9, but not TLR3 and TLR4, and that TLR4 deletion significantly reduced the expression of TLR1 and increased the expression of TLR9 (Fig. 2F).
TLR2, but not TLR4, Regulates the Self-renewal of in vitro-Expanded NPCs Isolated from the Postnatal Mouse Spinal Cord
We further investigated the role of TLR2 and TLR4 in the self-renewal and proliferation of in vitro-expanded NPCs isolated from the postnatal mouse spinal cord. Isolated and expanded WT, TLR2-/- or TLR4-/- NPCs all expressed Sox2 (Fig. 3A, representative images in orange; 3D, positive cells quantification) and FoxJ1 (Fig. 3A, representative images in green, 3C, positive cell quantification) to a similar degree (mRNA - Fig. 3B). We again found significant differences regarding Sox2 subcellular location – TLR4-/- NPCs displayed a significantly reduced number of cells with nuclear Sox2 expression compared to TLR2-/- NPCs or WT NPCs (Fig. 3D). This result agrees with the data found in Fig. 1H showing a significant reduction of the Sox2 positive cells at the perimeter SC.
We also evaluated the expression of marker genes preferentially expressed in immature NPCs (Notch1, Sox9, Dlx2, NCAM1, Olig1, and PDGFRα) in WT, TLR2-/-, or TLR4-/- NPCs to evaluate whether deletion of TLR2 or TLR4 could influence early glial (Sox9, Olig1, PDGFRα) or neuronal determination (Notch1, Dlx2, NCAM1). However, the mRNA expression analysis failed to find any difference in expression for the noted genes suggesting that TLR2 and TLR4 have not significantly influence on these early NPC multilineage markers at this early postnatal stage (Fig. 3E).
We next evaluated the ability of TLR2-/- and TLR4-/- NPCs to form primary neurospheres to explore their self-renewal capacity [42]. Overall, TLR2-/- NPCs formed significantly larger but less numerous neurospheres when compared to WT and TLR4-/- NPC-derived neurospheres (Fig. 3F), indicating the preferential formation of primary neurospheres and enhanced self-renewal. Meanwhile, TLR4-/- NPCs formed significantly smaller and more numerous neurospheres (Fig. 3F), indicating the more rapid formation of secondary neurospheres and limited self-renewal.
We also studied proliferation via BrdU incorporation and Ki67 immunostaining in WT, TLR2-/-, and TLR4-/- NPCs grown under adherent conditions, finding that only TLR2 deletion significantly increased NPC proliferation (Fig. 3G). Analysis of phospho-H2AX levels, which mark cells undergoing mitotic stress [43], revealed a significantly higher number of positive cells in TLR2-/- NPCs when compared to WT and TLR4-/- NPCs (Fig. 3H, representative images, right and graph, left). TLR2-/- NPCs also displayed a higher level of cMyc gene expression than TLR2-/- NPCs (Fig. 3I, left), indicative of enhanced cell cycle activity [44]. Analysis of p21 expression found significantly lower gene expression levels in TLR4-/- NPCs than WT NPCs, suggestive of a potentially deregulated cell cycle (Fig. 3I, right); however, PDL analysis suggested a slow-down in growth for TLR4-/- NPCs only (Fig. 3J).
Interestingly, both TLR2-/- and TLR4-/- NPCs exhibited increased apoptosis than WT NPCs. The significant increase in cell death could balance the increased proliferation in TLR2-/- NPCs to explain the lack of PDL differences compared to WT NPCs. As TLR4-/- NPCs did not display higher proliferative rates, the significantly increased apoptosis rate may explain the observed reduction in PDL (Fig. 3K).
Taken together, our data suggest that TLR4 maintains in vitro-expanded NPCs isolated from the postnatal mouse spinal cord in a proliferative and undifferentiated state., while TLR2 expression limits NPC proliferation and self-renewal.
TLR2 and TLR4 in Differentially Contribution to the Formation of Mature Neurons and Glial Cells
We finally sought to evaluate the contribution of TLR2 or TLR4 to glial or neuronal cell-fate determination by seeding WT, TLR2−/− or TLR4−/− NPCs onto Matrigel™ coated plates and analyzing comparing their spontaneous differentiation one day (undifferentiated) or seven days (spontaneously differentiated) after growth factor withdrawal, as indicated by the experimental schemes (Fig. 4A, B, top). We failed to encounter any differences in neuronal (β3Tubulin) (Fig. 4A, left panels, green) or astrocytic (GFAP) (Fig. 4A, right panels, red) differentiation of NPCs as a consequence of TLR2 or TLR4 loss after one day (undifferentiated cell condition); however, we encountered significant differences in neuronal (Fig. 4B, left panels) but not astrocytic (Fig. 4B, right panels) differentiation as a consequence of TLR2 or TLR4 loss after seven days. TLR2−/− NPCs displayed a highly significant increase in neuronal differentiation compared to WT and TLR4−/− NPCs, while TLR4−/− NPCs displayed significantly higher neuronal differentiation than WT NPCs (Fig. 4B, left panels).
Interestingly, the β3Tubulin-expressing neuron-like cells exhibited three different morphologies in all three genotypes (Fig. 4C). Type 1 cells are pyramidal-like neurons, with a prominent soma and numerous dendrites (a morphology compatible with mature neuron); Type 2 are immature-like neurons, which are small, with few neurites (a morphology compatible with undifferentiated neurons); and Type 3 cells were bipolar with very long axonal projections (a morphology compatible with mature neurons). TLR2−/− NPCs preferentially differentiated into Type 2 cells (immature neurons) and expressed elevated levels of Dcx, a marker of very early NPCs (Fig. 4D, upper graph). Meanwhile, TLR4−/− NPCs preferentially differentiated into Type 3 cells (mature neurons) but did not express higher levels of Map2, a marker of mature neurons (Fig. 4D, lower panel), perhaps indicating a transitional stage of maturation.
We next studied the neural cell fate identity of WT, TLR2−/− or TLR4−/− NPCs by analyzing Neurogenin1 expression levels [45]. Both self-renewing and spontaneously differentiated NPCs from TLR4−/− NPCs expressed higher levels of Neurogenin1) than WT and TLR2−/− NPCs (Fig. 4E), indicating a primed stage of NPCs for neuronal maturation. Overall, TLR2 loss prevented neuronal maturation, while TLR4 loss enhanced neuronal maturation.
We failed to find any differences in the percentage of astrocytic cells, based on the positive reactivity of GFAP, following analysis of WT, TLR2−/−, or TLR4−/− NPCs at one day or seven days after induced spontaneous differentiation (Fig. 4A, B); however, we discovered a significant increase in GFAP protein expression in TLR4−/− NPCs (Fig. 4F) previously associated to a reactive astrocytic phenotype [46]. Analysis of the Stat3 protein, previously described to be involved in astrocytic reactivity [47, 48], revealed a significant increase in TLR4−/− NPCs compared with WT or TLR2−/− NPCs (Fig. 4G), which could explain the reactive phenotypic profile found in the absence of TLR4.
Analysis of oligodendrocyte differentiation in suitable growth-supporting medium found that WT, TLR2−/− and TLR4−/− NPCs all expressed similar levels of Olig2, a transcription factor expressed in early to mature stage oligodendrocytes [49] (Fig. 4F, upper panels); however, after seven days of spontaneous differentiation, the absence of TLR2 significantly reduced the generation of Olig2 positive cells from NPCs (Fig. 4F, lower panels). We also found the significantly higher expression of Sox10, a transcription factor expressed in early OPCs [50], and the significantly lower expression of NG2, a factor expressed by mature OPCs [51], in the absence of TLR2 or TLR4. Overall, this data also suggests a critical role for TLR2 and TLR4 in oligodendrocyte maturation.