Sox2 and Mitf cross-regulatory interactions consolidate progenitor and melanocyte lineages in the cranial neural crest

Abstract

The cellular origin and molecular mechanisms regulating pigmentation of head and neck are largely unknown. Melanocyte specification is controlled by the transcriptional activity of Mitf, but no general logic has emerged to explain how Mitf and progenitor transcriptional activities consolidate melanocyte and progenitor cell fates. We show that cranial melanocytes arise from at least two different cellular sources: initially from nerve-associated Schwann cell precursors (SCPs) and later from a cellular source that is independent of nerves. Unlike the midbrain-hindbrain cluster from which melanoblasts arise independently of nerves, a large center of melanocytes in and around cranial nerves IX-X is derived from SCPs, as shown by genetic cell-lineage tracing and analysis of ErbB3-null mutant mice. Conditional gain- and loss-of-function experiments show genetically that cell fates in the neural crest involve both the SRY transcription factor Sox2 and Mitf, which consolidate an SCP progenitor or melanocyte fate by cross-regulatory interactions. A gradual downregulation of Sox2 in progenitors during development permits the differentiation of both neural crest- and SCP-derived progenitors into melanocytes, and an initial small pool of nerve-associated melanoblasts expands in number and disperses under the control of endothelin receptor B (Ednrb) and Wnt5a signaling.

Fig. 1.

Overview of melanocyte development during mouse embryonic development visualized using 3D projection images. Melanocyte development in relation to nervous tissues visualized by Sox10 (blue), Mitf (red) and NF (green) immunohistochemical staining. A, C, E and G show Sox10 and Mitf channels, and B, D, F and H show all three channels. (A,B) E9 embryo. Note nascent cranial ganglia V, VII-VIII, IX-X without any apparent Sox10⁺ NCC migration. (C-H) Note the emergence of Mitf⁺ cells in IX-X and hindbrain-midbrain clusters at E10 (C,D), their expansion at E10.5 (E,F) and dispersal at E11.0 (G,H). Scale bars: 500 μm.

Fig. 2.

Melanoblasts emerge in discrete and defined locations during mouse development. (A) Melanoblasts appear at E9.5 among Sox10⁺ cells associated with roots of cranial ganglia IX-X (boxed area and enlarged image in inset). (B) At E10, melanoblasts are detected in three distinct locations:; adjacent to cranial nerves IX-X (box 1), in the midbrain-hindbrain region (box 2) and in the anterior facial region of the head (box 3). (C-H) Higher magnification of boxed areas 1 (C,D), 2 (E,F) and 3 (G,H). Note the presence of Sox10⁺ cells in all three areas at E9 prior to emergence of Mitf⁺ cells at E10. (I-Q) E10.5-11 embryos. (I) Overview of locations of Mitf⁺ cells in the E10.5 embryo (boxed areas). (J-Q) Higher magnification images of boxed areas showing Mitf⁺ cells in: (J) roots and nerves of cranial ganglia IX-X, (K) vagal nerve (X) innervating embryonic gut, (L) oculomotor nerve (III), (M) central root of the trigeminal ganglion, (N) the facio-acoustic ganglion complex, (O,P) branches of facial nerve (VIII), (Q) dorsal part of the embryo at the forelimb level between neural tube and roots of DRGs. Arrows point to some Mitf⁺ cells. ov, otic vesicle. Roman numerals indicate the cranial nerves. Scale bars: 200 μm in A,B,I; 100 μm in J-Q.

Fig. 3.

Genetic tracing using PLP-CreERT/Rosa26-YFP mice and analysis of neuregulin receptor subunit (Erbb3) mutant mice reveal SCP-derived melanoblasts in cranial nerves IX-X. (A-C) Injection of TM at E9.5 and analysis at E10.5 of Mitf and YFP (A); Mitf, Tuj1, YFP (B); and Mitf, Sox10, YFP (C). Note the presence of YFP in a significant proportion of Mitf⁺ cells in the Tuj1⁺ and Sox10⁺ nerve and in Mitf⁺ cells associated with but not within the nerve. (D-F) Immunohistochemical staining for Cre recombinase, Mitf and YFP in PLP-CreERT2 mice at E10.5. Note the expression of Cre in SCPs of nerves, including those that are weakly Mitf⁺ within nerves (unfilled arrows), but not in strong Mitf⁺ cells (filled arrows). (G) Quantification of YFP⁺ cells in Sox10⁺ SCPs of nerves and in Mitf⁺ melanoblasts (n=4). (H) Quantification of Cre expression in Mitf⁺ cells in SCPs of nerves and melanoblasts outside cranial nerves IX-X. (n=4). (I,J) Whole-mount immunohistochemistry of E9.75 wild-type (I) and ErbB3^–/– (J) mouse embryos stained for NF (green) and Mitf (red). Arrows indicate clusters of melanoblasts associated with cranial nerves IX-X. (K,L) Cranial nerves IX-X of an E9.75 ErbB3^+/+ embryo. (M) Cross-section through cranial nerves IX-X of an E10.5 ErbB3^+/+ embryo. Note the numerous Sox10⁺ SCPs and Mitf⁺ melanoblasts associated with NF⁺ nerve fibers. (N,O) Cranial nerves IX-X of an E9.75 ErbB3^–/– embryo. (P) Cross-section through cranial nerves IX-X of an E10.5 ErbB3^–/– embryo. Note the reduction of Sox10⁺ SCPs and Mitf⁺ melanoblasts. (Q) Cell numbers of SCPs and melanocytes in wild-type and ErbB3^–/– E10.5 embryos (***P=0.001, n=4 embryos/genotype). (R) Percentage of melanoblasts (Sox10⁺:Mitf⁺) and SCPs (Sox10⁺:Mitf^–) relative to all Sox10⁺ cells in wild-type and ErbB3^–/– E10.5 embryos. Note that the proportion of Mitf⁺ cells is much higher in ErbB3^–/– than in wild-type embryos. Error bars represent s.e.m. Red line in H, Q and R separates analyzed tissues or genotypes. NT, neural tube; ov, otic vesicle. Roman numerals indicate the cranial nerves. Scale bars: (A-C) 50 μm inA-C; 30 μm in D-F; 250 μm in I,J; 100 μm in K,M,N,P; 50 μm in L,O.

Fig. 4.

Distinct roles of endothelin 3 for nerve IX/X and midbrain-hindbrain melanoblasts. (A-E) Cervical cluster of melanoblasts (A-D) and the whole head (E) of a wild-type control mouse embryo. Note that the cervical part of E10.5 embryo is also represented by 3D-rendered isosurfaces of cranial nerves IX-X (green) together with melanoblasts (orange) shown in transversal view in C and D. (F-J) Cervical cluster of melanoblasts (F-I) and the whole head (J) of an EDNRB^–/– embryo at E10.5. Note deficits of dispersal and expansion of melanoblasts associated with cranial nerves whereas only expansion is affected in midbrain-hindbrain melanocytes of EDNRB^–/– mutant embryos. (K) Quantification showing decreases of melanoblast numbers in the cervical cluster and scattered in the head compared with littermate controls at E10.5 (***P=0.0003, n=4 animal/genotype for cluster IX-X; **P=0.0011, n=4 animal/genotype for the whole head). Error bars represent s.e.m. (L-N) Whole-mount in situ hybridization with Edn3 riboprobe on mouse embryos at E9.5 (L), E10.5 (M) and E11 (N). Note the expression of Edn3 in otic vesicle (arrows). ov, otic vesicle. Roman numerals indicate the cranial nerves. Scale bars: 100 μm in A,B,F,G; 50 μm in C,D,H,I; 300 μm in E,J.

Fig. 5.

Sox2 and Mitf cross-regulatory interactions during melanocyte development in chick. (A) Cross-section through the forelimb level of E4.5 chick embryo stained for Sox2 (red) and Sox10 (blue). Boxed area is enlarged in inset. Note the lack of Sox2 expression in Sox10⁺ cutaneous melanocytes (arrows) and the gradual reduction of Sox2 in more distal parts of the ventral spinal nerve. (B,C) Mitf, Sox2 and Tuj1 staining of the brachial plexus. Sox2⁺ is observed in SCP whereas Mitf⁺ melanoblasts lack Sox2 (arrows). (D,E) Mitf, Sox2 and Tuj1 staining of the dorsal neural tube. Note the lack Sox2 immunoreactivity in the nuclei of NCC-derived Mitf⁺ cells (arrows) at E4.5. (F-I) NC overexpression of Sox2 and GFP in the chick. GFP labels overexpressing cells. In control GFP-overexpressing embryos, GFP labeled cells are observed among melanoblasts (Mitf⁺ cells) and the DRG (arrows). Sox2 expression leads to loss of GFP⁺/Mitf⁺ cells (white arrows) and ectopic ganglia-like formations (yellow arrows). (J-L) Inducible overexpression of Sox2 and GFP at E5.0 in the chick with analysis 12 hours later. Non-overexpressing melanoblasts (i.e. GFP^– cells) contain Mitf (unfilled arrows) whereas GFP⁺ cells completely lack Mitf immunoreactivity (filled arrows). (M-O) Inducible overexpression of Mitf for 6 hours caused significant repression of Sox2 in targeted (GFP⁺) cells (arrows) in CNS and PNS, and weak ectopic Mebl1 expression in spinal cord. N and O are high magnification images of the boxed area in M. (P-R) Inducible overexpression of Mitf for 12 hours caused a complete loss of Sox2 in targeted cells in CNS (unfilled arrows) and PNS (filled arrows) (P). Ectopic expression of Mebl1 was observed in overexpressing (GFP⁺) cells (Q). Dotted line outlines the neural tube. BP, brachial plexus; SN, ventral spinal nerve; NT, neural tube. Scale bars: 150 μm in A; 70 μm in A, inset; 50 μm in B,C; 100 μm in D-R.

Fig. 6.

Sox2 regulatory effects on melanocyte development in NCC conditionally mutant mice (Wnt1Cre/Sox2^fl/fl). (A-D) Expression of Sox2 (blue) and Mitf (red) in cells associated with cranial nerve IX in E10.5 mouse embryos. White arrows point at Mitf⁺ cells with no or very low levels of Sox2 (C,D), yellow arrows show Mitf⁺ cells retaining low levels of Sox2 (C,D). (E) Sox2 expression (red) decreases proximo-distally in the ventral spinal nerve (SN) in the E12 mouse embryo. (F,G) Sox2 expression in SCPs of spinal nerve dorsal rami. Arrows indicate Sox10⁺ prospective melanoblasts under the epidermis that are negative for Sox2. (H) Schematic model of Sox2 and Mitf cross-repressive interactions during development of the glial and melanocyte lineages. (I-L) Conditional deletion of Sox2 in the NCCs using the Wnt1-Cre activator strain. Melanoblasts numbers at the cranial nerves IX-X cluster markedly increased in conditional null mice (K,L) compared with control mice (I,J). (M,N) Magnified view of part of J and L. (O) Quantification of Mitf⁺ cells in IX-X cluster of wild-type and Wnt1Cre/Sox2^fl/fl mice at E10.5 (left graph, P<0.0001, n=4). (P) Quantification of all Sox10⁺ cells and Sox10⁺/Mitf^– cells in cranial nerves IX/X of Wnt1Cre/Sox2^fl/fl and wild-type embryos (for Sox10⁺/Mitf^– cells **P=0.0096, n=4 embryos/genotype). (Q) Percentage of Mitf⁺/Sox10⁺ cells among all Sox10⁺ cells between Wnt1Cre/Sox2^fl/fl embryos and littermate wild-type embryos (**P=0.0012, n=4 embryos/genotype). (R) Quantification of melanoblasts appearing adjacent to cranial ganglia V and VII/VIII in Wnt1Cre/Sox2^fl/fl embryos and littermate control embryos (ganglion V: *P=0.0161, n=3 Wnt1Cre/Sox2^fl/fl and n=5 in control; ganglia VII/VIII: **P=0.0014, n=4 Wnt1Cre/Sox2^fl/fl and n=8 in control). (O-R) Error bars represent s.e.m. NT, neural tube; ov, otic vesicle; SN, ventral spinal nerve. Roman numerals indicate the cranial nerves. Scale bars: 100 μm in A,B,F,G,M-P; 50 μm in C,D; 500 μm in I-L.

Fig. 7.

Sox2 regulates the activity of Mitf-m proximal promoter. (A) Structure of Mitf-m proximal promoter with predicted binding sites. Arrows show the positions of primers used to amplify site A and site B in ChIP assay with Sox2 antibody. (B) PCR reaction demonstrating binding of Sox2 on the proximal Mitf-m promoter in ChIP assay. (C) Domain structure of Sox2 and Sox2-activating and -repressing fusion proteins used for overexpression experiments in vitro and in vivo. (D) Luciferase reporter assay in melanoma cells with endogenous Mitf-m promoter activity (control, n=16 vs: Sox2, n=8, ***P<0.0001; Sox2-ENR construct, n=8, ***P<0.0001; Sox2-VP16 construct, n=4, **P=0.0038). (E-L) In vivo effects of Sox-ENG and Sox2-VP16 on melanocyte development 2 days after electroporation in chick. Arrows in E,F,I,J show GFP⁺/Mitf⁺ migratory cells and in G,H, all GFP⁺ cells. Solid arrows show NCC-derived cells, whereas open arrows point to cells in the neural tube. (K,L) Dashed line separates two halves of a neural tube. Note the presence of nuclear Mitf (red) in GFP⁺ cells of the electroporated half. NT, neural tube. Scale bars: 100 μm.