Non-canonical Targets of HIF1a Impair Oligodendrocyte Progenitor Cell Function

Abstract

Mammalian cells respond to insufficient oxygen through transcriptional regulators called hypoxia-inducible factors (HIFs). Although transiently protective, prolonged HIF activity drives distinct pathological responses in different tissues. Using a model of chronic HIF1a accumulation in pluripotent-stem-cell-derived oligodendrocyte progenitors (OPCs), we demonstrate that HIF1a activates non-canonical targets to impair generation of oligodendrocytes from OPCs. HIF1a activated a unique set of genes in OPCs through interaction with the OPC-specific transcription factor OLIG2. Non-canonical targets, including Ascl2 and Dlx3, were sufficient to block differentiation through suppression of the oligodendrocyte regulator Sox10. Chemical screening revealed that inhibition of MEK/ERK signaling overcame the HIF1a-mediated block in oligodendrocyte generation by restoring Sox10 expression without affecting canonical HIF1a activity. MEK/ERK inhibition also drove oligodendrocyte formation in hypoxic regions of human oligocortical spheroids. This work defines mechanisms by which HIF1a impairs oligodendrocyte formation and establishes that cell-type-specific HIF1a targets perturb cell function in response to low oxygen.

Keywords: HIF1a; OPCs; differentiation; hypoxia; oligodendrocyte progenitor cells.

Figure 1.. HIF1a Accumulation Impairs the Induction of Oligodendrocytes from OPCs

(A) Western blot of HIF1a in sgVhl and Cas9 control OPCs. (B) qRT-PCR of Vegfa and Bnip3 in Cas9 control (in blue) and sgVhl (in green) OPCs. Data are presented as mean ± SEM from 3 biological replicates. (C) Western blot of HIF1a in hypoxic (1% O₂) and normoxic OPCs. (D) qRT-PCR of Vegfa and Bnip3 in hypoxic (in green) and normoxic (in blue) OPCs. Data are presented as mean ± SEM from 3 biological replicates. (E) Schematic of the two in vitro differentiation schemes directing OPCs to either oligodendrocytes or astrocytes. (F) Representative images of Cas9 control and sgVhl oligodendrocytes (MBP+ in green) and astrocytes (GFAP+ in green). Nuclei are marked by DAPI (in blue). Scale bar, 100μm. (G) Quantification of the percentage of oligodendrocytes (MBP+ cells / DAPI) and astrocytes (GFAP+ cells / DAPI) formed from sgVhl (in green) and Cas9 control OPCs (in blue). Data are presented as mean ± SD from 3 independent biological replicates. p-values were calculated using Student’s two-tailed t-test. (H) Schematic illustrating acquisition of early (O4), intermediate (O1) and late (MBP) oligodendrocyte markers during in vitro oligodendrocyte differentiation. (I) Representative images of early (O4+ in green), intermediate (O1+ in green) and late (MBP+ in green) oligodendrocytes during days 1, 2, and 3 of differentiation of Cas9 control and sgVhl OPCs. Nuclei are marked by DAPI (in blue). Scale bars, 100μm. J-L) Quantification of the percentage of early O4+ (J), intermediate O1+ (K), and late MBP+ (L) oligodendrocytes in sgVhl (in green) and Cas9 control OPCs (in blue) at days 1, 2, and 3 of differentiation. Data are presented as mean ± SD from 3 independent biological replicates. p-values were calculated using two-way ANOVA (reported as ANOVA p=) for overall group differences and Sidak’s multiple comparisons test for individual timepoint differences. p-values for qRT-PCRs were calculated using Student’s two tailed t-test. See also Figure S1.

Figure 2.. HIF1a directly upregulates unique targets in OPCs.

(A) Aggregate binding profile and heatmap of 503 HIF1a peaks (FDR<0.001) in sgVhl OPCs within 2Kb of the transcription start site (TSS) and transcription end site (TES) of the closest expressed gene in Cas9 control and sgVhl OPCs. See also Table S1 for full list of peaks. (B) Venn diagram of direct HIF1a target genes in diverse mouse tissues overlapped with HIF1a targets in OPCs. See also Table S2 for full list of peaks. (C) Volcano plot of genes that significantly increase and decrease in yellow and blue respectively (P-adj < 0.05) in sgVhl OPCs compared to Cas9 control OPCs with direct targets of HIF1a in black. Gray dots are genes not significantly different between conditions. Data are from 3 biological replicates. (D) Violin plots of expression data (TPM) normalized to Cas9 control OPCs for canonical and OPC-specific HIF1a targets in Cas9 control (in blue) and sgVhl (in green) OPCs. Bold dashed line represents the median with the thin dashed lines representing the upper and lower quartiles. p-values were calculated using the Mann-Whitney test. (E) Gene ontology (GO) analysis of shared canonical targets of HIF1a. Table shows the rank of the GO term along with −log(p-value). (F) Heatmap of row normalized expression of glycolysis genes (TPM) between Cas9 control and sgVhl OPCs. Genes in the gray box are direct targets of HIF1a. Each row represents a biological replicate. (G) Gene ontology (GO) analysis of OPC-specific HIF1a target genes. Table shows the rank of the GO term along with −log(p-value). See also Figure S2.

Figure 3.. Chromatin Accessibility and Cell-Type-Specific Transcription Factors Define Non-Canonical HIF1a Targets.

(A) Aggregate plots of HIF1a and H3K27ac enrichment in Cas9 control and sgVhl OPCs and open chromatin enrichment in non-transduced OPCs at shared canonical HIF1a peaks, OPC-specific HIF1a peaks, and “other tissue-specific” HIF1a peaks. (B) Heatmap of the enrichment of the top 10 transcription factor motifs under non-canonical HIF1a peaks in OPCs compared to other tissue types. Olig2 (highlighted in red) is a dominant transcription factor of the oligodendrocyte lineage. (C) Aggregate plots of OLIG2 enrichment in Cas9 control and sgVhl OPCs at canonical HIF1a peaks, OPC-specific HIF1a peaks, and “other tissue-specific” HIF1a peaks. D) Western blot of OLIG2 from total cell lysate (input), IgG control lysate, and lysate from pulldown (IP) of either OLIG2 (on left) or HIF1a (on right) in Cas9 control (top) and sgVhl OPCs (bottom). See also figures S3H and S3I for reciprocal co-IP and replicate co-IP in sgVhl OPCs. (E) Genome browser view of H3K27Ac, HIF1a, and OLIG2 ChIP-seq in Cas9 control (blue) and sgVhl (green) OPCs at Bnip3 and Ascl2. Scale bars, 5 Kb and 2 Kb respectively. See also Figure S3.

Figure 4.. OPC-specific HIF1a Targets Ascl2 and Dlx3 Suppress Sox10 and Impair Oligodendrocyte Formation.

(A) Quantification of the normalized expression (TPM) of oligodendrocyte lineage markers in Cas9 control (in blue) and sgVhl (in green) OPCs. Data represent mean ± SEM from 3 independent RNA-seq replicates. (B) Genome browser view of H3K27ac and HIF1a ChIP-seq in Cas9 control (in blue) and sgVhl (in green) OPCs at the locus for Sox10. The gray bar highlights significant H3K27Ac enrichment at the Sox10 promoter in control cells (−log₁₀(q value) = −5.6) that is absent in Vhl knockout OPCs. Scale bar, 2Kb. (C) qRT-PCR of Sox10 and downstream SOX10 target genes Pdgfra and Plp1 in sgVhl (in green) and Cas9 control (in blue) OPCs. Data are presented as mean ± SEM from 3 biological replicates. (D) Western blot for SOX10 in sgVhl, Vhl.2 OPCs, and Cas9 control OPCs. Data represent results from a single biological replicate. (E) Venn diagram overlapping genes with proximal HIF1a binding (in green), significant gains in H3K27ac (in blue), and significantly increased expression (in purple) in OPCs. The top 10 genes were then ranked and listed by their enrichment score, which is the product of the Log₂(Fold change in gene expression) and −Log(P-adj). See also Table S1 for the full target list. (F) qRT-PCR of 8 top HIF1a target genes (see Figure 1D for Vegfa and Bnip3) in hypoxic (1% O₂, in green) and normoxic (21% O₂, in blue) OPCs. Data are presented as mean ± SEM from 3–4 technical replicates. (G) Genome browser view of HIF1a ChIP-seq at Ascl2 and Dlx3 in sgVhl OPCs (in green), Cas9 control OPCs (in blue), and other diverse cell types. H3K27ac and OLIG2 enrichment are also shown in sgVhl and Cas9 control OPCs. The gray bars highlight HIF1a accumulation in sgVhl OPCs. Scale bars, 2Kb. (H) Quantification of the percentage of early O4+, intermediate O1+, and late MBP+ oligodendrocytes from sgAscl2 (in yellow), sgDlx3 (in dark blue) and sgControl (in light blue) CRISPRA OPCs at day 1, 2 and 3 of differentiation. Data are presented as mean ± SD of 6–8 technical replicates (individual wells) per condition. (I) qRT-PCR of Sox10 in sgAscl2 (in yellow), sgDlx3 (in dark blue), and sgControl (in light blue) CRISPRA OPCs. Data are presented as mean ± SEM from 3 biological replicates. (J) Genome browser view of HA ChIP-seq upstream of Sox10 in Cas9 control (in blue) and HA-Ascl2 (in yellow) OPCs. The gray bars highlight enhancers for Sox10 defined by H3K27Ac ChIP-seq in Cas9 Control OPCs. Scale bar, 5 Kb. p-values for qRT-PCRs were calculated using Student’s two-tailed t-test. See also Figure S4.

Figure 5.. Chemical Inhibition of MEK/ERK Increases Oligodendrocyte Formation from sgVhl OPCs

(A) Primary bioactives library screen showing the effect of 1753 molecules on percentage of oligodendrocytes (MBP+ cells/ total DAPI) formed by sgVhl OPCs relative to DMSO treated sgVhl OPCs. The dotted line represents a 3-fold increase in oligodendrocyte formation and compounds that clear this threshold are indicated as green dots. MEK inhibitors are highlighted as gray boxes and pro-myelinating drugs are highlighted as yellow dots. See also Table S5. (B) Pie charts of the number of MEK inhibitors (in dark gray) and other compounds classes (in blue) within top hits compared to their prevalence in the non-toxic compounds of the Selleck library. p-value was calculated using hypergeometric analysis. (C) Representative immunocytochemistry images of oligodendrocytes (MBP+ in green) from the primary drug screen of the top 5 MEK inhibitor hits along with the DMSO negative control. Nuclei are marked by DAPI (in blue). Scale bars, 100μm. (D) Heatmap showing the row normalized fold change in the percentage (MBP+ / DAPI) of oligodendrocytes relative to DMSO treated sgVhl OPCs of an 8-point dose curve of the primary hits. The heatmap rows are sorted by unsupervised hierarchical clustering with columns in order from high (10μM) to low dose (78nM). MEK inhibitors are highlighted in gray and bolded. Data are presented as the mean from 3 separate dose curve plates. See also Table S6. (E) Averaging all 8 doses shows the ability of each MEK inhibitor to increase the formation of oligodendrocytes (MBP+/DAPI) relative to DMSO treated sgVhl OPCs. Green and blue columns represent the most and least effective compounds respectively. Data are presented as the mean ± SD from 3 separate dose curve plates. p-values were calculated using a one-way ANOVA with Tukey’s multiple comparisons test. (F) Representative western blot for phosphorylated ERK1/2 (p-ERK1/2) relative to total ERK1/2 of sgVhl OPCs incubated with 100nM of indicated MEK inhibitors for 30 minutes. (G) Quantification of the ratio of p-ERK1/2 to total ERK1/2 for the most (in green) and least (in blue) effective MEK inhibitors relative to DMSO treated controls. Data are presented as mean ± SD from 3 biological replicates. p-values were calculated using one-way ANOVA with Dunnett’s multiple comparisons test. See also Figure S5.

Figure 6.. MEK Inhibitors Restore Sox10 Expression Independent of HIF1a Activity in OPCs

(A) Western blot for HIF1a in sgVhl OPCs treated with AZD8330 or DMSO. Data represent results from a single biological replicate. (B) Violin plot of normalized expression (TPM values normalized to Cas9 control+DMSO) of direct functional targets of HIF1a (see Figure 2B) in Cas9 control+DMSO (in blue), sgVhl + DMSO (in green), and sgVhl + AZD8330 (in purple) OPCs. (C) Quantification of normalized expression (TPM) for both Ascl2 and Dlx3 in Cas9 control+DMSO (in blue), sgVhl+DMSO (in green) and sgVhl+AZD8330 (in purple) OPCs. Data represent mean ± SD from 3 independent RNA-seq replicates. (D) Gene set enrichment analysis (GSEA) analysis of gene program changes in sgVhl compared to Cas9 control OPCs demonstrates a significant reduction (FDR<0.001) in GO terms for Oligodendrocyte Development (normalized enrichment score/NES = −2.74) and Oligodendrocyte Differentiation (NES = −3.11). (E) GSEA analysis of gene program changes in sgVhl + AZD8330 compared to sgVhl + DMSO OPCs demonstrates a significant enrichment (FDR<0.001) in GO terms for Oligodendrocyte Development (normalized enrichment score/NES = 2.98) and Oligodendrocyte Differentiation (NES = 3.58). (F) Violin plot showing normalized expression (TPM values normalized to Cas9 control + DMSO OPCs) of genes in the GO term Oligodendrocyte Development (GO:0014003) that decrease (FC<0.75) in sgVhl + DMSO OPCs (in green) relative to Cas9 control + DMSO OPCs (in blue) as well as sgVhl OPCs following treatment with AZD8330. (G) qRT-PCR of Sox10 and Myrf in Cas9 control+DMSO (in blue), sgVhl+DMSO (in green) and sgVhl+AZD8330 (in purple) OPCs. Data are presented as mean ± SEM from 4 technical replicates (individual wells). (H) Schematic of human brain oligocortical spheroids treated at days in vitro (DIV) 70 with either DMSO or AZD8330 for 4 days. At day 90, organoids were incubated with hypoxyprobe, fixed, and sectioned for immunohistochemistry. (I) Representative immunohistochemistry images for oligodendrocytes (MYRF+ in red) and hypoxic regions (hypoxyprobe in green) of DIV 90 oligocortical spheroids at low and high magnification that were treated from DIV 70–74 with DMSO or AZD8330. Scale bars, 100μM. (J) Quantification of oligodendrocytes (MYRF+ / mm²) in the whole oligocortical spheroid (total), hypoxic region of the spheroid (hypoxyprobe positive), and normoxic region of the spheroid (hypoxyprobe negative) in DIV 90 spheroids that had been treated with DMSO or AZD8330 from DIV 70–74. Data represent mean ± SD from 8–9 individual spheroids. p-values were calculated using Student’s two-tailed t-test. AZD8330 was used at a dose of 300nM for all experiments and p-values for violin plots were calculated using the Kruskal Wallis One-Way ANOVA with Dunn’s multiple comparisons test. See also Figure S6.