Ubiquitin ligase COP1 coordinates transcriptional programs that control cell type specification in the developing mouse brain.

The E3 ubiquitin ligase CRL4COP1/DET1 is active in the absence of ERK signaling, modifying the transcription factors ETV1, ETV4, ETV5, and c-JUN with polyubiquitin that targets them for proteasomal degradation. Here we show that this posttranslational regulatory mechanism is active in neurons, with ETV5 and c-JUN accumulating within minutes of ERK activation. Mice with constitutive photomorphogenesis 1 (Cop1) deleted in neural stem cells showed abnormally elevated expression of ETV1, ETV4, ETV5, and c-JUN in the developing brain and spinal cord. Expression of c-JUN target genes Vimentin and Gfap was increased, whereas ETV5 and c-JUN both contributed to an expanded number of cells expressing genes associated with gliogenesis, including Olig1, Olig2, and Sox10. The mice had subtle morphological abnormalities in the cerebral cortex, hippocampus, and cerebellum by embryonic day 18 and died soon after birth. Elevated c-JUN, ETV5, and ETV1 contributed to the perinatal lethality, as several Cop1-deficient mice also lacking c-Jun and Etv5, or lacking Etv5 and heterozygous for Etv1, were viable.

Constitutive photomorphogenesis 1 (COP1; also called RFWD2) is the evolutionarily conserved substrate adaptor of the cullin-RING ubiquitin ligase CRL4COP1/DET1 (1). COP1 binds to De-etiolated 1 (DET1), which in turn binds to Damage-specific DNA binding protein 1 (DDB1) that is in complex with scaffold protein CUL4A and Ring-box 1 (2–4). Substrates of mammalian CRL4COP1/DET1 include the transcription factors c-JUN, ETV1, ETV4, ETV5, ETS1, ETS2, and C/EBPα, the metabolic enzyme acetyl-CoA carboxylase, and CREB regulated transcription coactivator 2 (2, 3, 5–8). Ubiquitination of these substrates targets them for proteasomal degradation.

Studies with COP1-deficient mice suggest that regulation of c-JUN, ETV1, and ETV4 abundance contributes to tumor suppression by COP1 (3, 9). For example, COP1 deficiency in mouse prostate epithelial cells results in elevated ETV1, ETV4, and c-JUN and early prostate intraepithelial neoplasia (3). COP1 also has an important role in regulating ETV1, ETV4, and ETV5 levels in pancreatic β-cells so that insulin secretion is not perturbed (10). CRL4COP1/DET1 activity is tightly regulated, with phosphorylation of DET1 by ERK contributing to inactivation of the ligase. Consequently, cells exposed to growth factors exhibit rapid accumulation of CRL4COP1/DET1 substrates (4).

Although COP1 deficiency in many tissues is deleterious, inappropriate CRL4COP1/DET1 activity can also promote disease. For example, C/EBPα degradation orchestrated by COP1 in combination with the pseudokinase TRIB1 (or TRIB2) is linked to impaired myeloid differentiation and the development of acute myeloid leukemia (7, 11). Inappropriate CRL4COP1/DET1 activity may also have consequences for the development of the central nervous system because increased COP1 copy number is linked to autism in humans (12).

Potential substrates of CRL4COP1/DET1 in the brain include the transcription factors c-JUN, ETV1, ETV4, and ETV5. Activator protein-1 (AP-1) transcription factor c-JUN is broadly expressed in the developing brain, regulates neuronal apoptosis in response to phosphorylation by upstream kinases (13), and is a substrate of both CRL4COP1/DET1 and the ubiquitin ligase SCFFBW7. Posttranslational regulation of c-JUN is critical to neurogenesis because deletion of Fbw7 from neural stem cells elevates c-JUN and reduces cell viability (14). Regulation of c-JUN abundance by FBW7 in the granule cell layer also plays an important role in development of the cerebellum (15).

ETS family members ETV1, ETV4, and ETV5 have several roles in the developing nervous system (16–20), but less is known about their posttranslational regulation in the brain. In developing mouse cortex, Etv1 mRNA marks a subpopulation of early Cajal–Retzius neurons that are specified by FGF8 signaling, whereas, at later stages, it serves as a marker of layer five cerebrocortical neurons (21–23). Etv4 and Etv5 are also expressed early in cortical development, but they show a distinct pattern of expression to Etv1 (23, 24). Expression of Etv5 in the ventricular zone of the mouse cerebral cortex is induced by the MAPKs MEK1 (also called MAP2K1) and MEK2 (MAP2K2), and this is proposed to confer an astrocytic fate on neural stem and progenitor cells (18). MEK1 and MEK2 are part of the RAF–MEK–ERK kinase cascade that is engaged by RAS GTPases. Consistent with this pathway regulating the switch from neurogenesis to gliogenesis, deletion of the RAS negative regulator neurofibromatosis 1 (NF1) from neural stem cells promotes ERK-dependent gliogenesis at the expense of neurogenesis in the olfactory bulb during perinatal stages (19). ETV transcription factors have also been implicated in glioma initiation by oncogenic RAS (25). Whether ERK-dependent posttranslational mechanisms, potentially involving CRL4COP1/DET1, regulate expression of ETV1, ETV4, and/or ETV5 protein in the brain has not been examined to our knowledge. We investigated how COP1 impacts brain development by deleting Cop1 in neural stem and progenitor cells with Nestin.cre transgenic mice, or in cells of the neocortex and hippocampus with Emx1 knock-in mice.

Results

COP1 Mediates Posttranslational Regulation of c-JUN, ETV1, and ETV5 During Brain Development.

To determine whether COP1 and DET1 expressed in the developing mouse brain interact with known CRL4COP1/DET1 substrates, epitope-tagged versions of COP1 and DET1 were affinity-purified from embryonic day 18.5 (E18.5) knock-in mouse brains (Fig. 1). ETV5 and DET1 copurified with Flag-HA-COP1. These interactions were specific because ETV5 and COP1 also copurified with DET1-3xFlag, but not with an unrelated protein, ARMC8-3xFlag. Consistent with COP1 being the substrate adaptor for ETV5, deletion of Cop1 from neural stem cells with a Nestin.cre transgene (Cop1ΔN mice) reduced the amount of COP1 and ETV5 that copurified with DET1-3xFlag, but had no effect on the interaction of DET1 with DDB1.

Fig. 1.

Posttranslational regulation of ETV5 and c-JUN in neurons. (A) Western blots of E18.5 mouse brain lysates after immunoprecipitation (IP) with anti-FLAG M2 beads. (B) Western blots of cortical neurons cultured for 2 d and then stimulated with 100 ng/mL BDNF with or without 1 μM ERK inhibitor (ERKi) for 2 h. mRNA expression was determined by quantitative RT-PCR. (C) Western blots of cortical neurons cultured for 14 d and then stimulated with 100 μM PTX with or without 1 μM ERKi for 5 min. (D) Western blots of cortical neurons infected with adeno-associated viruses expressing WT or mutant COP1. Results in A–D are representative of three independent experiments.

One consequence of posttranslational regulation of protein stability is that protein abundance may not correlate with mRNA abundance. Evidence for posttranslational regulation of ETV5 was obtained with E15.5 cortical neural cultures, which we treated with BDNF or picrotoxin (PTX) to mimic the signals that neurons might encounter during development. BDNF engages the tyrosine kinase receptor TrkB on neurons, whereas PTX blocks the GABA-activated chloride channel and thereby promotes synaptic activity. BDNF treatment for 2 h increased ETV5 protein and Etv5 mRNA, as well as c-JUN protein and c-Jun mRNA (Fig. 1). ERK inhibition, which reduced phosphorylation of the ERK substrate ribosomal S6 kinase, prevented this response to BDNF. However, PTX increased ETV5 and c-JUN protein abundance in an ERK-dependent manner within 5 min without increasing Etv5 and c-Jun mRNA expression (Fig. 1). This rapid accumulation of ETV5 and c-JUN is consistent with them being subject to posttranslational regulation. Interestingly, ectopic COP1 prevented the PTX-induced increase in ETV5, whereas mutant COP1Δ24 that is unable to bind to DET1 (2) had no effect (Fig. 1). This result suggests that ERK-mediated inhibition of CRL4COP1/DET1 is inefficient when COP1 is in excess.

We sought genetic proof that COP1 regulates ETV5 in the developing mouse brain by using Cop1ΔN (Cop1 Nestin.cre) mice and Cop1ΔE (Cop1 Emx1 ) mice. Many Cop1ΔN mice died within a few days of birth, and none survived to weaning (). Abnormalities were evident in the postnatal day 0 (P0) cerebral cortex, hippocampus, and cerebellum (Fig. 2). Neurons in cortical layers 2 and 3 of the frontal and parietal regions appeared more densely packed than in littermate controls, the thickness of the molecular layer in the hippocampus was reduced, and the granular cell layer of the cerebellum exhibited indistinct lobulation and hypocellularity. Cop1ΔE mice, which lacked COP1 in cells of the neocortex and hippocampus (), displayed similar cerebrocortical disorganization as Cop1ΔN mice (). Cop1ΔE mice had a median survival of 147 d (), and those that survived to 35 d tended to be smaller than their Cop1 Emx1 littermates ().

Fig. 2.

COP1 is required for normal brain development. Brain sections from P0 littermates stained with H&E. Boxed regions in columns 1 and 3 are shown at higher magnification in columns 2 and 4, respectively. (Scale bars: 50 μm; boxed regions, 200 μm.) Results are representative of four mice per genotype.

In keeping with COP1 regulating ETV5 abundance via a posttranslational mechanism, E18.5 or P0 Cop1ΔN brains contained more ETV5 than control Cop1 Nestin.cre brains (Fig. 3 ), but they did not express more Etv5 mRNA (Fig. 3 ). CRL4COP1/DET1 substrates c-JUN and ETV1 were also more abundant in Cop1ΔN brains (Fig. 3 and ), even though c-Jun and Etv1 mRNAs were not increased (Fig. 3 and ). Indeed, Cop1ΔN brains contained less Etv1 or Etv5 mRNA than control brains (Fig. 3), suggestive of a negative feedback signaling loop in response to elevated ETV1 and ETV5. ETV1, ETV5, and c-JUN were also more abundant in Cop1ΔE brains (). We confirmed the specificity of our ETV1 and ETV5 monoclonal antibodies with ETV1- and ETV5-deficient mouse brains ().

Fig. 3.

COP1 limits expression of ETV1, ETV5, and c-JUN during brain development. (A) Western blots of P0 mouse brains. (B) E18.5 cerebral cortices labeled for ETV5 (brown), Etv5 (white), or COP1 (brown). (Scale bars: 200 μm.) IHC, immunohistochemistry; ISH, in situ hybridization. Results are representative of two mice per genotype. (C) Relative Etv1, Etv5, and c-Jun mRNA expression in E18.5 brains. Circles represent individual embryos. (D) E14.5 cerebral cortices labeled for ETV5 (red) or c-JUN (brown). (Scale bars: ETV5, 100 μm; c-JUN, 200 μm.). IF, immunofluorescence. (E) E13.5 spinal cords labeled for ETV1 (brown) or ETV5 (green) with c-JUN (red). (Scale bars: ETV1, 50 μm; ETV5/c-JUN, 100 μm.) Results in D and E are representative of three to four mice per genotype.

Given that Cop1, Etv1, Etv5, and c-Jun are also expressed in the brain and spinal cord at earlier stages of embryogenesis () (16, 23, 26) and that cre activity in Nestin.cre mice is detected by E11, we examined Cop1ΔN brains between E12.5 and E16.5 to see when we could first detect elevated levels of CRL4COP1/DET1 substrates. Labeling of Cop1ΔN brains at E12.5 was variable, but, at E14.5, the Cop1ΔN cerebral cortex exhibited more intense c-JUN staining and contained significantly more ETV5+ cells than control cortex (Fig. 3). COP1 deficiency also increased staining for ETV1, ETV5, and c-JUN in the E13.5 spinal cord, particularly surrounding the central canal (Fig. 3). Cop1ΔN brains were indistinguishable morphologically from control brains at these earlier stages of embryogenesis.

Similar to what we observed in whole brain, cortical neural cultures from E16.5 Cop1ΔN embryos contained more ETV5 protein than control cultures, even though Etv5 mRNA was not increased (). In addition, BDNF treatment did not further increase ETV5 in the Cop1ΔN cells, suggesting that the BDNF-induced increase in ETV5 in control cells could partly reflect ERK-mediated inactivation of COP1.

Combined ETV5 and c-JUN Deficiency Reduces Lethality in Cop1ΔN Mice.

To determine the contribution of abnormally high c-JUN, ETV1, and ETV5 protein expression to the lethal phenotype of Cop1ΔN mice, we introduced floxed c-Jun, Etv1, and Etv5 alleles (27–29). In addition, we confirmed that each of the three transcription factors was increased independent of the others in Cop1ΔN brains (Fig. 4). Interestingly, E18.5 Cop1/c-JunΔN brains differed from Cop1ΔN brains because they exhibited abnormally elevated expression of ETV5 in the periventricular region of the cerebrum (Fig. 4). Etv5 mRNA expression in the Cop1/c-JunΔN periventricular region appeared intermediate between the level seen in control Cop1 Nestin.cre brains and the reduced level seen in the Cop1ΔN brains (). These data suggest that c-JUN negatively regulates Etv5 expression in the periventricular region, although whether this is direct or indirect regulation is unclear. c-JUN deficiency alone also increased ETV5 expression in the periventricular region ().

Fig. 4.

Negative regulation of Etv5 by c-JUN and of Etv4 by ETV1 and ETV5. (A) Western blots of E18.5 mouse brains. tr. ETV5, truncated ETV5. (B) E18.5 cerebral cortices labeled for Pax6 (green) and ETV5 (red). (Scale bars: 50 μm.) Results are representative of three mice per genotype. (C) Western blots of E18.5 brains.

Cop1/c-JunΔN mice, like Cop1ΔN mice, did not survive to weaning (). However, a number of Cop1/Etv5ΔN mice and Cop1/c-Jun/Etv5ΔN mice were weaned. Most of the Cop1/Etv5ΔN mice died soon thereafter, but approximately one third of the Cop1/c-Jun/Etv5ΔN mice appeared healthy at 3 mo of age (). Deletion of Etv1 with Nestin.cre was lethal around weaning, similar to what has been reported for Etv1− mice (16), so we could only halve the Etv1 gene dosage. Etv1 heterozygosity did not increase the proportion of Cop1/c-Jun/Etv5ΔN mice alive at weaning, but it allowed some Cop1/Etv5ΔN mice to survive for many months, similar to Cop1/c-Jun/Etv5ΔN mice (). We conclude from these data that aberrant c-JUN, ETV1, and ETV5 protein expression, although not entirely responsible, contributes to the lethality of Cop1ΔN mice.

Complete rescue might not have been achieved because loss of ETV5 alone or in combination with ETV1 loss increased expression of the related transcription factor ETV4 (Fig. 4). This finding is reminiscent of combined ETV4 and ETV5 deficiency causing ETV1 expression in pancreatic β-cells lacking COP1 (10). Such observations suggest that ETV1, ETV4, and ETV5 negatively regulate their own expression. Consistent with this notion, Etv4 expression was increased in E18.5 Cop1/Etv1/Etv5ΔN brains, particularly in periventricular cells, compared with COP1ΔN brains ().

COP1 Deficiency Enhances Expression of Genes Associated with Gliogenesis in a c-JUN/ETV5–Dependent Manner.

To determine the gene-expression changes caused by aberrant expression of c-JUN, ETV1, and ETV5, we analyzed E18.5 control (Cop1 Nestin.cre), Cop1ΔN, Cop1/c-JunΔN, Cop1/c-Jun/Etv1ΔN, and Cop1/c-Jun/Etv1/Etv5ΔN brains by RNA sequencing [complete data available at the Gene Expression Omnibus (GEO) database, accession no. GSE111564]. Compared with controls, 97 genes were dysregulated in Cop1ΔN brains (based on an adjusted P value <0.05 and fold change >1.5). Of these 97 genes, 57 seemed to be regulated by c-JUN, ETV1, and/or ETV5 because their expression reverted to control levels in Cop1/c-Jun/Etv1/Etv5ΔN brains. Within this subset of 57 genes, Gfap, Vimentin, Olig1, Olig2, Sox10, and Cspg4 were up-regulated 1.6–4.0-fold in the Cop1ΔN brains and stood out because of their involvement in gliogenesis (30–33).

By quantitative RT-PCR, increased expression of Gfap or Vimentin in E18.5 Cop1ΔN brains was dependent on c-JUN because Gfap and Vimentin expression in Cop1/c-JunΔN brains was equivalent to that in control brains (Fig. 5). These data are consistent with reports that Gfap and Vimentin are c-JUN target genes (34, 35). In contrast, enhanced Olig1, Olig2, and Sox10 expression in Cop1ΔN brains was sustained in Cop1/Etv5ΔN brains, partially normalized in Cop1/c-JunΔN brains, and completely normalized to control levels in Cop1/c-Jun/Etv5ΔN brains (Fig. 5 ). We speculate that elimination of both c-JUN and ETV5 is needed to normalize expression of Olig1, Olig2, and Sox10 because c-JUN loss increased ETV5 in the periventricular region of Cop1ΔN brains (Fig. 4).

Fig. 5.

c-JUN/ETV5-dependent changes in gene expression in COP1-deficient brains. (A) Relative Olig1, Olig2, Sox10, Gfap, and Vimentin expression in E18.5 brains by quantitative RT-PCR. Circles represent individual embryos. (B) E18.5 cerebral cortices labeled for Olig1 mRNA (purple). (Scale bars: 50 μm.) Results are representative of three mice per genotype. (C) E18.5 cerebral cortices labeled for OLIG2 (green) and ETV5 (red). (Scale bars: 100 μm.) Graph shows quantification of labeling. Circles represent individual embryos.

Immunofluorescence labeling of E18.5 Cop1ΔN cerebral cortex revealed that more cells expressed OLIG2 protein (Fig. 5), which correlated well with the increase in Olig2 mRNA expression in whole brain (Fig. 5). OLIG2-expressing cells were also more abundant in E18.5 Cop1ΔE cortex (). OLIG2 and ETV5 were largely expressed in different cell populations, although a small number of periventricular cells showed overlapping expression in Cop1/c-JunΔN brains (Fig. 5). Note that the Etv5 mutant allele in Cop1/Etv5ΔN brains encodes truncated, transcriptionally inactive ETV5 (29), and we detect this protein with our ETV5 antibody.

Sustained RAS–MEK–ERK signaling in neural stem cells lacking NF1 induces ectopic OLIG2 expression (19), and ERK signaling inactivates CRL4COP1/DET1 (4), so we wondered whether COP1 deficiency enhanced gliogenesis at the expense of neurogenesis in the cerebral cortex. However, E18.5 control and Cop1ΔN cerebral cortices contained comparable numbers of cells expressing the neuronal marker NeuN (). In addition, even though we detected a 1.7-fold increase in Cspg4 expression in E18.5 Cop1ΔN whole brain by RNA sequencing, cells expressing CSPG4/NG2 (chondroitin sulfate proteoglycan 4; also called neuron-glial antigen 2) were not increased in the Cop1ΔN cortex by immunohistochemistry (). It is possible that the subtle increase in Cspg4 expression in E18.5 Cop1ΔN whole brain reflects differences outside the cerebral cortex. Colabeling of NeuN and OLIG2 in control and Cop1ΔE brains from littermates aged 5 wk indicated a trend toward increased OLIG2+ cells, whereas NeuN+ cell numbers appeared unchanged (). These data suggest that COP1 deficiency was not inducing a simple fate switch such that gliogenesis was favored at the expense of neurogenesis.

To further validate these findings, we performed single-cell RNA sequencing on E16.5 cerebral cortices isolated from Cop1ΔN and control brains (; complete data available at the GEO database, accession no. GSE111704). Consistent with our earlier results, the Cop1ΔN cortex contained a greater proportion of cells expressing Olig2, Olig1, or Sox10 (Table 1). These cells were assigned to similar clusters as control cells expressing Olig2, Olig1, or Sox10, and the level of gene expression per cell was comparable between the two genotypes (, clusters 7, 8, and 12). These data suggest that the extra Olig2-expressing cells in the Cop1ΔN cerebral cortex represent the expansion of a normally occurring cellular subset rather than a population of cells aberrantly expressing Olig2. For example, the Cop1ΔN cortex did not contain a preponderance of cells aberrantly coexpressing Olig2 and the neuronal marker gene Microtubule associated protein 2 (Map2; ). The Cop1ΔN cortex also contained approximately fourfold more cells expressing Gfap, whereas the increase in the number of cells expressing Vimentin was minimal (Table 1). Consistent with COP1 deficiency not reducing NeuN+ neurons in the E18.5 cerebral cortex (), numbers of immature neurons expressing Doublecortin (Dcx) or postmitotic neurons expressing Map2 were comparable between E16.5 control and Cop1ΔN cortices (Table 1). Therefore, markers of neurogenesis in the embryonic cerebral cortex do not appear to be suppressed by COP1 deficiency despite aberrant expression of markers of gliogenesis.

Table 1.

Single-cell RNA sequencing of E16.5 cortical cells

Gene	Positive (%)
	Cop1fl/+ Nestin.cre	Cop1fl/− Nestin.cre (Cop1ΔN)
Olig1	124 (1.4)	419 (4.2)
Olig2	186 (2.1)	513 (5.1)
Sox10	40 (0.4)	88 (0.9)
Dcx	7,227 (80.8)	7,616 (76.0)
Map2	5,763 (64.4)	6,076 (60.6)
Vim	2,867 (32.1)	3,749 (37.4)
Gfap	68 (0.8)	330 (3.3)

Cells expressing the genes listed in column 1 are enumerated and shown as percentages of all cortical cells analyzed. Data represent cortical cells from two mice of each genotype. A gene is considered expressed if the expression value in that cell is greater than zero.

Given that OLIG2-expressing cells were increased in E18.5 Cop1ΔN cerebral cortex in a cJUN- and ETV5-dependent manner, we explored how E16.5 cells expressing c-Jun or Etv5 clustered. c-Jun expression was detected across all cell clusters (), in keeping with c-JUN protein being broadly expressed in the developing brain (Fig. 3 and ). However, as expected, Etv1, Etv4, and Etv5 showed more restricted patterns of expression (). All three RNAs were detected in clusters 8 and 12, but we also noted expression in clusters 7 (Etv1), 14 (Etv1, Etv4), and 16 (Etv1, Etv5). Although Olig2-expressing cells also occupied clusters 7, 8, and 12, they did not appear to coexpress Etv1, Etv4, or Etv5 (), consistent with distinct cell populations expressing ETV5 and OLIG2 protein (Fig. 5). Confirming our earlier results (Fig. 3 and ), Cop1ΔN cells expressed less Etv1, Etv4, and Etv5 than control cells (). Collectively, our results suggest that stabilization of c-JUN and/or ETV1/4/5 in a subset of cells lacking COP1 promotes the expansion of a population of cells expressing glial marker genes. Determination of the precise relationship between these cell populations will require further study.

Discussion

Complete loss of COP1 is deleterious to the developing mouse embryo around E9.5 (9), so we explored the role of COP1 in brain development by restricting Cop1 deletion to neural stem and progenitor cells with a Nestin.cre transgene. These Cop1ΔN mice died soon after birth with morphological abnormalities in the cerebral cortex, hippocampus, and cerebellum (Fig. 2). Isolated heart and kidney cells of Nestin.cre mice also exhibit cre activity, but brain abnormalities in the Cop1ΔN mice probably caused the perinatal lethality because 22% (22 of 101) of Cop1ΔE mice also died before weaning, compared with 5% (6 of 125) of Emx1 littermate controls.

The precise cause of death of Cop1ΔN or Cop1ΔE mice was unclear, but aberrant expression of ETV5, c-JUN, and ETV1 contributed to the Cop1ΔN phenotype because several Cop1 Etv5 Etv1 Nestin.cre mice and Cop1/c-Jun/Etv5ΔN mice survived for 7 mo. Abnormally elevated expression of ETV4 in the absence of ETV5 (Fig. 4) might explain why not all mice were rescued. Consistent with ETV1 and ETV5 suppressing Etv1, Etv4, and Etv5 gene expression, increased amounts of ETV1 and ETV5 in the Cop1ΔN brains coincided with reduced expression of all three genes (Fig. 3 and ). It is unclear if this represents direct transcriptional repression by ETV1 and ETV5. c-JUN may also suppress Etv5 gene expression because ETV5 protein and Etv5 mRNA were more abundant in the periventricular region of the Cop1/c-JunΔN cerebrum compared with Cop1ΔN brains (Fig. 4 and ). A c-JUN binding site is detected between exons 6 and 7 of Etv5, albeit without an underlying AP-1 motif (36). Regardless of the exact mechanism, suppression of Etv1, Etv4, or Etv5 gene expression by ETV1/4/5 or c-JUN appears to be a negative feedback loop for limiting the ETV1/4/5 transcriptional response.

Neighboring ETS and AP-1 transcription factor binding sites are recognized RAS-response elements (37). Both ETV1 and ETV4 of the ETS family are capable of binding to these genomic DNA sequences (38). Therefore, in cells expressing c-Jun and one or more of Etv1, Etv4, and Etv5, CRL4COP1/DET1 inactivation by the RAS–MEK–ERK pathway (4) could be a mechanism for the rapid and coordinate accumulation of c-JUN and ETV1/4/5 for binding to RAS-response elements. Indeed, ERK-dependent expression of the transcription factor c-MAF in the lens of the developing mouse eye has been linked to c-JUN and ETV5, which bind to ETS-AP1 sites in the c-Maf locus and synergistically activate transcription in reporter studies (39). This synergy might reflect simultaneous binding of ETV1/4/5 and AP-1 to the MED25 subunit of the mediator transcriptional coactivator complex that engages RNA polymerase II (40).

The genes that were up-regulated in E18.5 Cop1ΔN whole brain in a c-JUN/ETV5–dependent manner included Olig1, Olig2, and Sox10 (Fig. 5), which are markers of oligodendrocyte precursor cells. Single-cell RNA sequencing of E16.5 cerebral cortex () confirmed OLIG2 immunohistochemistry experiments (Fig. 5) and showed that more Cop1ΔN cells expressed these genes (Table 1). Given that Cop1ΔN cells expressing Olig1, Sox10, or Olig2 fell into similar clusters as their control counterparts (), these data suggest that COP1 deficiency promotes expansion of a normal cellular subset rather than promoting aberrant gene expression in cells that do not normally express Olig1, Sox10, or Olig2. Cells expressing markers of neurogenesis, including NeuN, Dcx, and Map2, were unchanged () or minimally reduced in the Cop1ΔN cortex (Table 1), indicating that gliogenesis was not enhanced at the expense of neurogenesis. Perhaps COP1 loss enhanced proliferation and/or reduced apoptosis in a progenitor population that was already committed to a glial fate. In contrast to neurogenesis in the embryonic cerebral cortex, neurogenesis in the olfactory bulb perinatally is perturbed by enhanced RAS–MEK–ERK signaling, with NF1 deficiency skewing the differentiation of neural stem cells in the subventricular zone toward the glial lineage and yielding a smaller olfactory bulb at P18 (19). Because of the lethal phenotype of newborn Cop1ΔN mice, we could not determine if COP1 loss mimicked NF1 deficiency in this setting.

Finally, our study highlights the fact that measuring Etv1, Etv4, and Etv5 mRNA expression in cells of the developing brain is insufficient to implicate these transcription factors in normal physiology because ligases such as CRL4COP1/DET1 can prevent the accumulation of functional amounts of ETV1, ETV4, and ETV5 protein.

Materials and Methods

Armc8, Cop1, Cop1, Cop1, c-Jun, Det1, Etv1, Etv5, Nestin.cre, and Emx1 mouse strains have been described previously (3, 4, 27–29, 41–43). Embryos were designated E0.5 on the morning that a vaginal plug was observed. Pups were defined as P0 on the day of birth. Images in Fig. 1 used cortical cells from E15–E16 CD-1 mouse embryos (Charles River Laboratories). The Genentech Animal Care and Use Committee approved all animal protocols. Detailed methods and reagents are provided in .