Sox5 functions as a fate switch in medaka pigment cell development

Abstract

Mechanisms generating diverse cell types from multipotent progenitors are crucial for normal development. Neural crest cells (NCCs) are multipotent stem cells that give rise to numerous cell-types, including pigment cells. Medaka has four types of NCC-derived pigment cells (xanthophores, leucophores, melanophores and iridophores), making medaka pigment cell development an excellent model for studying the mechanisms controlling specification of distinct cell types from a multipotent progenitor. Medaka many leucophores-3 (ml-3) mutant embryos exhibit a unique phenotype characterized by excessive formation of leucophores and absence of xanthophores. We show that ml-3 encodes sox5, which is expressed in premigratory NCCs and differentiating xanthophores. Cell transplantation studies reveal a cell-autonomous role of sox5 in the xanthophore lineage. pax7a is expressed in NCCs and required for both xanthophore and leucophore lineages; we demonstrate that Sox5 functions downstream of Pax7a. We propose a model in which multipotent NCCs first give rise to pax7a-positive partially fate-restricted intermediate progenitors for xanthophores and leucophores; some of these progenitors then express sox5, and as a result of Sox5 action develop into xanthophores. Our results provide the first demonstration that Sox5 can function as a molecular switch driving specification of a specific cell-fate (xanthophore) from a partially-restricted, but still multipotent, progenitor (the shared xanthophore-leucophore progenitor).

Figure 1. Larval pigment pattern phenotypes of medaka ml-3 mutants.

(A–H) 9 dpf (hatching stage). (A, C, E, G) WT. (B, D, F, H) ml-3 mutant. (E, F) Bright field. (G, H) UV light. (A–D) Dorsal views. (E–H) Lateral views. (A–D) Leucophores are formed in excess in mutants. Whereas in WT larvae leucophores lie over the head (A) and along dorsal midline of the trunk (C) in association with melanophores, in ml-3 larvae excess leucophores are localized ectopically to form bilateral lines along the anterior-posterior axis (B, D) in addition to those in the normal position in head (B) and trunk midline (D). (E–H) Xanthophores are absent from ml-3 mutants. In WT (E) as well as in ml-3 (F), xanthophores are not evident when observed in brightfield. Under UV illumination, however, auto-fluorescent xanthophores are readily visible in WT, covering over dorsal trunk (G). In ml-3 mutants, xanthophore auto-fluorescence is not detected (H). (A–H) Melanophore patterns are unaffected in ml-3 mutants. In both WT (E, G) and ml-3 mutants (F, H), melanophores are distributed in three longitudinal stripes (DS, LS and VS). (I, J) Counts of leucophores and melanophores in ml-3 mutants at 9 dpf. Mean counts (error bars give standard deviation) are shown for WT (blue) and ml-3 mutants (red). (I) The numbers of leucophores in head region and in dorsal trunk are larger in ml-3 than in WT (n = 12 for each group and region): two-way ANOVA, F(group) = 389.09, df = 1,42, p<0.0001; F(group×region) = 46.93, df = 1,42, p<0.0001 and Student's t-test for each region, *, p<0.0001. (J) There is no significant difference in melanophore number between WT and ml-3 (n = 12 for WT, n = 11 for ml-3): two-way ANOVA, F(group) = 3.72, df = 1,88, p>0.05; F(group×region) = 0.74, df = 1,88, p>0.05 and Student's t-test for each region, **, p>0.05. (K, L) Mean (±s.d.) counts of leucophores and xanthophores on dorsal trunk surface in WT, ml-3 heterozygotes (+/ml-3) and ml-3 homozygotes (ml-3) at 5 dpf (also see Figures S2E–S2J). (K) The leucophore numbers seem to be negatively correlated with the copy-number of ml-3 gene (n = 10 for each group). p<0.0001 by one-way ANOVA. ***, p<0.0001 by Student's t-test for WT versus ml-3 heterozygotes, WT versus ml-3 homozygotes, and ml-3 heterozygotes versus homozygotes. (L) The xanthophore numbers seem to be positively correlated with the copy-number of ml-3 gene (n = 10 for each group). p<0.0001 by one-way ANOVA. ****, p<0.0001 by Student's t-test. Leu, leucophore; Xan, xanthophore; DS, dorsal stripe; LS, lateral stripe; VS, ventral stripe. Scale bars: 250 µm.

Figure 2. Embryonic pigment cell precursor formation in ml-3 mutants.

(A, B, D, F′, G′, H, I), gch. (K, L), dct.(M, N), mitfa. (E, E′) gch (blue) and dct (red). (A, C–H, K, M) WT. (B, I, L, N) ml-3 mutant. (A, B, E, F, H, I, K–N) Lateral views. (C, D, G) Dorsal views. (H′, I′) Transverse sections. (C, F, G) Pre-fixed samples in darkfield. (A–D) 27 somite stage (27 s, 61 hpf). (E, H, I, K–N) 34 somite stage (34 s, 74 hpf). (F, F′, G, G′) 7 dpf. (A–I) ISH analyses show gch expression in WT xanthophores and leucophores. Whereas gch-expressing cells are found over the lateral and anterior trunk surface and the head in WT at 27 s (A) as well as in ventral head leucophores (black arrowheads, C, D), ml-3 embryos have gch expression only in ventral head leucophores (black arrowheads, B). The embryo at 34 s is co-stained with gch riboprobe (purple) and dct riboprobe (red) (E, E′). gch signals show no overlap with dct signals. At 7 dpf, gch expression is detected on the surface of the whole length of trunk in WT (F′, G′). The dotted gch signals in the dorsal midline coincide with leucophore positions (compare F′ with F and G′ with G, some examples were represented by white arrowhead in G and G′). In WT at 34 s (H), gch-expressing cells are spread more posteriorly and scattered over the dorsal trunk surface (H′). In ml-3 mutants (I), gch expression is in fewer cells (J), and these are restricted to the dorsal trunk surface (I′) compared with WT. Transverse histological section from embryos at the level as indicated by dotted line in H and I. (J) Counts of gch-expressing cells at 34 s (n = 11 for WT, n = 10 for ml-3). The number is significantly fewer in ml-3 than in WT (*, p<0.0001 by Student's t-test). (K–N) Melanophore precursors, detected using dct (K, L) and mitfa (M, N) are not markedly altered in ml-3 (L, N) as compared with WT (K, M). (O) The number of dct- or mitfa-expressing cells is not significantly different between WT and ml-3 (dct, n = 11 for each group, **, p>0.05 by Student's t-test; mitfa, n = 12 for each group, **, p>0.05 by Student's t-test). (J, O) Mean (±s.d.) counts are shown as bars in blue (WT) and in red (ml-3). Scale bars: (A, F, H, K) 250 µm; (C) 100 µm; (E) 150 µm; (H′) 25 µm.

Figure 3. Mapping of the ml-3 locus.

(A) ml-3 locus was mapped to a 90 kb region (red bar) between MLG23-1 and MLG23-2 on LG23, predicted to contain only one gene, sox5. Typing markers are indicated above the map, with the number of recombinants per total 720 haploid genomes examined at each position. (B) Medaka sox5 comprises 15 exons (upper). Sequencing of cDNAs showed that in ml-3, exon 7 is skipped (lower). Boxes represent exons. Angled lines represent introns. The 5′ and 3′ untranslated regions are colored in black. Diagonal stripes and colored regions correspond to regions encoding the protein domains described in (D). Black arrows show positions of the primer set used for RT-PCR in C. (C) RT-PCR detects the skipping of exon 7 in sox5 mRNA of ml-3 mutant. (D) WT sox5 gene encodes 742 amino acid protein consisting of two coiled-coil domains (diagonal stripe) and HMG box (red box). The first coiled-coil domain contains a leucine-zipper (green box) and a glutamine-rich domain (Q box, light green box). In ml-3, loss of exon 7 causes a premature stop codon leading to a truncated protein. Gray box represents an altered frame. The resultant truncated protein would lack the second coiled-coil domain and HMG box (lower).

Figure 4. sox5 morphant and allelic TILLING mutants show the ml-3 pigment phenotype.

(A–D) Injection of sox5 MO into WT embryo generates ml-3 mutant phenocopies with increased leucophores in the dorsal trunk compared with embryos injected with the control MO which all show a WT leucophore phenotype (A). The severity of the leucophore phenotype is classified into “weak” (B) and “severe” (C). In “weak” embryos, the majority of leucophores are located along the midline of the trunk, while fewer leucophores are ectopically positioned (B). “Severe” embryos show a phenotype indistinguishable from that of ml-3 (C). In most cases, general morphology is normal, although about half of sox5 morphants combined a “severe” pigment phenotype with many ectopic leucophores with gross morphological abnormalities (data not shown). (D) Using this phenotypic classification, we scored the severity of the ml-3 morphant phenotypes at three different doses (1 ng, 2 ng and 4 ng) and with control MO (con). The number of injected embryos is indicated above the bars (n). (E) Our TILLING screen for sox5 mutant alleles identified two distinct mutations causing amino acid substitutions at N541S (blue box) or F543I (green box) respectively in the HMG domain (red boxed). (F–K) Homozygotes for the TILLING mutations (sox5^N541S and sox5^F543I) are compared with WT siblings (F, G). These sox5 mutants exhibit ml-3 mutant phenocopies as manifested in ectopic and excessive formation of leucophores and complete absence of xanthophores. Leucophores are shown in the darkfield image (H, J). Xanthophores are not detectable under UV light (I, K). Scale bars: 500 µm.

Figure 5. Expression pattern of medaka sox5 in WT and ml-3 mutant embryos.

(A, C, E, F) WT. (B, D) ml-3 mutant. (A′, B′, C″, D″) Transverse sections. (E, F) sox5 (blue) and gch (red). (A–E, E′) Lateral views. (F, F′) Dorsal views. (A) In WT embryos at 12 somite stage (12 s, 41 hpf), sox5 is expressed in premigratory NCCs (see also section in A′, black arrow) as well as in dorsal neural tube and CNS ranging from forebrain (fb) to hindbrain (hb), and in tailbud (tb). The boundary between neural tube and somite is indicated by dotted line. (B) In ml-3, the sox5 expression pattern is not markedly altered. In particular, when observed in section (B′), premigratory NCCs are positive for sox5. (C, C′, C″) At 24 somite stage (24 s, 58 hpf), sox5-expressing cells are found in dorsal neural tube, premigratory NCCs (arrows in C″) and migrating NCCs between neural tube and somite and lateral trunk surface (black arrowheads in C′, C″) pathways in WT. sox5-expressing cells scattered on lateral trunk surface are prominent in WT (C′, C″). (D, D′, D″) In ml-3, sox5 expressing cells are absent from lateral trunk surface (D′, D″), whereas sox5 expression remains in dorsal neural tube, premigratory NCCs (black arrows) and migrating NCCs between neural tube and somite (D″). Boxed portion in C, D are magnified in C′, D′, respectively. (A′, B′, C″, D″) Transverse histological section from embryos at the level as indicated by dotted line in A, B, C′ and D′. (E, E′) sox5-expressing cells on lateral trunk surfaces at 34 somite stage (34 s, 74 hpf) also express gch. White arrowhead represents an example of gch-positive sox5-expressing cell. (F, F′) On dorsal trunk surface, some sox5-negative gch-positive cells were detected (white arrows). Scale bars: (A, C) 200 µm; (C′) 100 µm; (A′, C″) 20 µm; (E) 50 µm.

Figure 6. Cell transplantation.

(A, C, E, F, H, I) Transplants of WT donor cells into ml-3 hosts. (B, D, G, J) Transplants of ml-3 mutant donor cells into WT hosts. (A–D) Note that leucophores are fluorescent through both UV and GFP filters (white arrowheads) and xanthophores are fluorescent through UV filter but not through GFP filter. In both types of transplant experiments, xanthophores and leucophores express GFP when generated from donor cells having slc2a15b:GFP transgene, but in leucophores the GFP signal is masked by strong auto-fluorescence of these cells. However, GFP expression is readily detectable by immunostaining (see Figures S5A–S5C). (A, C) In WT→ml-3 transplants, ectopic leucophores are partially lost and instead a patch of xanthophores (marked with dotted circle) positive for UV (A) and GFP (C) fluorescence can be seen. (B, D) In ml-3→WT transplants, leucophores and xanthophores develop and are positioned normally although leucophores derived from donor cells cannot be identified because GFP fluorescence is masked by strong auto-fluorescence. No xanthophores show GFP fluorescence (compare UV image, B and GFP image, D). (E, H) In one of WT→ml-3 transplants, a few GFP positive leucophores are found along midline (black arrows). (F, I) In another WT→ml-3 transplants, leucophores are ectopically formed and positive for GFP (black arrowheads). White arrowheads represent donor-derived xanthophores detected by anti-GFP antibody. (G, J) In ml-3→WT transplants, GFP-positive leucophores develop and are positioned normally along midline (black arrows). All images are dorsal views at 7 dpf. Xan, xanthophore; Leu; leucophore. Scale bars: 100 µm.

Figure 7. Dual expression of sox5 and pax7a is required for xanthophore development.

(A, B) gch. (C–E, I′) pax7a. (F, G, H′) sox5. (A, F) lf mutant. (B, E, G) lf-2 mutant. (C, H, I) WT. (D) ml-3 mutant. (A–G) Lateral views. (H, I) Dorsal views. (C′, C″, D′, D″, E′, F′, G′) Transverse section. (A, B) gch expression in lf and lf-2 at 34 somite stage (34 s, 74 hpf). gch is normally expressed in lf (A), whereas in lf-2 gch-expressing cells are completely absent (B). (C–E) pax7a expression in WT, ml-3 and lf-2 at 24 somite stage (24 s, 58 hpf). In WT, pax7a is expressed in dorsal neural tube, premigratory NCCs (black arrows) in migratory NCCs between neural tube and somite and on lateral trunk surface (black arrowheads, C–C″). In ml-3, pax7a-expressing cells are lost from lateral trunk surface (D–D″). In lf-2, pax7a expression in premigratory and migratory NCCs was absent (E, E′). (F, G) sox5 expression in lf and lf-2 at 24 s. sox5 expression is normal in lf being detected in dorsal neural tube, premigratory NCCs (black arrows) and migrating NCCs between neural tube and somite and on lateral trunk surface (black arrowheads) as in WT (F, F′). In lf-2, premigratory NCC expression remains detectable, but sox5-expressing cells are lost from lateral trunk surface (G, G′). (E′–G′) The histological section from embryos at the level as indicated by dotted line in E, F and G. (C′, C″, D′, D″) Transverse sections from embryos were selected from the region boxed by dotted line in C and D. (H, I) Expression of sox5 and pax7a on dorsal trunk surface in WT at 90 hpf. (H, I) Pre-fixed WT embryos in bright field. The same embryos were processed for sox5 (H′) and pax7a (I′) ISH. Leucophores are positive for pax7a expression (I′, compare white arrowhead positions with I) but not for sox5 expression (H′). Scale bars: (A, C, E) 200 µm; (C′, E′) 20 µm; (G) 50 µm.

Figure 8. Model for xanthophore and leucophore development from neural crest.

We propose that xanthophores and leucophores develop from shared progenitors. Sox5 functions to control fate specification of xanthophores in place of leucophores. In the progenitors, which are positive for pax7a, sox5-expressing cells are specified to xanthophore fate whereas pax7a-expressing sox5-negative cells give rise to leucophores. In ml-3 mutants, loss of functional Sox5 causes a failure of xanthophore specification, resulting in all progenitors becoming specified to leucophore fate. The phenotypes of lf-2 (pax7a) and ml-3 (sox5), which are independent of melanophore and iridophore lineages, suggest that xanthophore and leucophore share common progenitors. Mel, melanophore; Iri, iridophore; Xan, xanthophore; Leu, leucophore.