A regulatory sub-circuit downstream of Wnt signaling controls developmental transitions in neural crest formation

Abstract

The process of cell fate commitment involves sequential changes in the gene expression profiles of embryonic progenitors. This is exemplified in the development of the neural crest, a migratory stem cell population derived from the ectoderm of vertebrate embryos. During neural crest formation, cells transition through distinct transcriptional states in a stepwise manner. The mechanisms underpinning these shifts in cell identity are still poorly understood. Here we employ enhancer analysis to identify a genetic sub-circuit that controls developmental transitions in the nascent neural crest. This sub-circuit links Wnt target genes in an incoherent feedforward loop that controls the sequential activation of genes in the neural crest lineage. By examining the cis-regulatory apparatus of Wnt effector gene AXUD1, we found that multipotency factor SP5 directly promotes neural plate border identity, while inhibiting premature expression of specification genes. Our results highlight the importance of repressive interactions in the neural crest gene regulatory network and illustrate how genes activated by the same upstream signal become temporally segregated during progressive fate restriction.

Fig 1. Chromatin conformation capture identifies a tissue-specific enhancer that interacts with the AXUD1 promoter.

(A-B) Whole mount in situ hybridization for AXUD1, depicting the specific mRNA expression during neural crest specification stages HH8 (A) and HH9 (B). (C) Schematic representation of a Chromosome Conformation Capture-qPCR (3C-qPCR) experiment. Crosslinked neural crest cells were incubated with restriction enzyme NCOI and a DNA ligase. These steps allow the formation of hybrid DNA molecules combining restriction fragments that were in close proximity in nuclei of the cells. Primers spanning the AXUD1 locus were paired with a primer anchored in the AXUD1 promoter to amplify hybrid DNA junctions and quantify the interaction frequency with the promoter. (D-F) Identification of active enhancers in the AXUD1 locus. 3C-qPCR interaction map for the AXUD1 locus reveals regions of high interaction frequency with the promoter region (blue dotted line, D). Gray dotted lines highlight the six elements tested in transient transgenesis assays (see below). Error bars represent ± SEM. Purple and blue lines in (D) represent two replicates of the same 3C experiment. ATAC-seq, H3K27ac and TFAP2A CUT&RUN profiles at AXUD1 locus depict regions of accessibility and active chromatin regions (E). (F) eRNA quantification (RT-PCR, normalized to reference gene) for the regions numbered in (E) indicates the level of transcription in the promoter region and putative distal regulatory elements. Error bars represent ± SEM. Statistical significance determined via an unpaired t-test. (G) Reporter vector used in transgenesis reporter assays. The construct consists of the candidate enhancer region cloned upstream of the HSV-tk minimal promoter driving eGFP expression. (H-J) The Axud1E1 element is active in neural crest cells. In vivo activity of Axud1E1 as shown by eGFP expression in reporter assays (H). Axud1E1 recapitulates endogenous gene expression in dorsal neural folds (arrows), while the putative enhancer ax.1 displayed no specific activity (I). Transverse cryosection from a HH10 embryo electroporated with the Axud1E1 enhancer illustrates eGFP expression in the dorsal neural tube and migratory neural crest cells (J). HH, Hamburger and Hamilton. Scale bars represent 500μm (A-B), 200μm (H-I) and 100μm (J). *p < 0.05, **p < 0.01.

Fig 2. Dissection of the Axud1E1 enhancer.

(A-D) Transient transgenesis expression pattern of Axud1E1 depicting robust activity during neural crest specification (A), and in pre-migratory (B), early (C) and late migratory neural crest cells (D). (E-F) Double fluorescent in situ hybridization for AXUD1 (magenta) and eGFP (Axud1E1, turquoise) in transgenic embryos shows colocalization of the endogenous gene (E) and the enhancer reporter (F). (G) ATAC-seq profile at Axud1E1 and enhancer dissection strategy. Turquoise bars represent enhancer variants that were able to generate strong reporter activity in transgenic embryos. Blue and gray bars represent weak and inactive enhancer variants, respectively. (H-I) Dorsal view of embryos transfected with constructs containing Axud1E1-500 and Axud1E1-SE (Axud1E1 shadow enhancer) defined in (G). Axud1E1-500 displays robust activity in neural crest cells (arrows in H). The Axud1E1-SE region displays weaker activity in the same cell population (arrows in I). (J) Quantification of fold change in Axud1E1-500 and Axud1E1-SE eRNA levels when Axud1E1-500 is targeted with specific gRNAs. Error bars represent ± SEM. Statistical significance determined via an unpaired t-test. HH, Hamburger and Hamilton. Scale bars represent 200μm (A-D, E-F, H-I). *p < 0.05, **p < 0.01.

Fig 3. Axud1E1 is directly regulated by nuclear effectors of canonical Wnt signaling.

(A-C) Axud1E1 responds to Wnt pathway manipulation. Dorsal view of an embryo electroporated with control (left) and WNT1/4 (right) morpholinos (A) and representative images showing the loss of Axud1E1-500 activity upon WNT1/4 (B) or CTNNB1 (C) knockdown in the right side of the embryos. (D) Quantification of the effect of WNT1/4 (n = 12) and CTNNB1 (n = 15) loss-of-function assays on Axud1E1. Statistical significance determined via ANOVA. (E-G) Whole embryo in situ hybridization for TCF7 (E), TCF7L2 (F) and LEF1 (G) shows that LEF1 is the only Wnt nuclear effector robustly expressed in cranial regions during neural crest specification and early migration stages (arrows). (H) Immunohistochemistry for LEF1 depicting enrichment of the protein in migrating neural crest cells (arrows). (I-K) Wnt effectors CTNNB1 and LEF1 directly bind to Axud1E1. Chromatin immunoprecipitation for LEF1, TCF7, TCF7L1 and TCF7L2, performed with neural folds of WT embryos, shows that LEF1 is the only Wnt nuclear effector that co-immunoprecipitates Axud1E1. Interaction between LEF1 and Axud1E1 (I) was not detected in neural plate border tissue dissected from gastrula stage embryos (HH4), and was lost upon treatment with a Wnt1 dominant negative (DN) construct (J). Axud1E1-SE did not interact with LEF1 (J). ChIP with a CTNNB1 antibody also revealed robust interaction with Axud1E1 (K). Error bars represent ± SEM. HH, Hamburger and Hamilton; MO, morpholino; NF, neural fold; DN, dominant negative. Scale bars represent 200μm (A-C, E-H). ***p < 0.001.

Fig 4. Mutation analysis of Axud1E1 reveals putative upstream regulators.

(A) ATAC-seq and TFAP2A occupancy profiles at Axud1E1. The blue horizontal bar describes the 100bp regions of Axud1E1-500 mutated in this analysis. (B-F) Transient transgenesis of Axud1E1-500 mutants shown in (A). Mutations in regions A (B) and B (C) do not eliminate enhancer activity (arrows). In contrast, enhancer activity in neural crest cells is drastically reduced in mutants C (D), D (E) and E (F). (G-H) Co-expression strategy for enhancer-reporter quantification. Transient transgenic embryo co-electroporated with the control Axud1E1 cloned into the vector pTK-mCherry (G) and the enhancer variant cloned into the pTK-eGFP vector (H). Cranial region of transgenic embryos were dissected and 500 Cherry+ cells were analyzed with flow cytometry. (I) Categorical scatterplot depicting the reporter intensity of Axud1E1-500 mutant constructs as measured by flow cytometry. Statistical significance determined via an unpaired two-tailed t-test. (J) Transient transgenesis shows activity of an Axud1E1-500 variant in which both regions A and B were mutated. (K) Transient transgenesis of Axud1E1-300 shows robust reporter expression in neural crest cells (arrows). (L) Categorical scatterplot depicting reporter intensity in Axud1E1, Axud1E1-500 and Axud1E1-300. The statistical significance was determined via an unpaired two-tailed t-test. (M) Sequence of Axud1E1-300 highlighting the MSX1, ZIC1 and SP5 binding motifs defined using JASPAR database. Kb, kilobase; HH, Hamburger and Hamilton; Scale bars represent 100μm (B-H, J-K); ***p < 0.001; ****p < 0.0001; ns, not significant.

Fig 5. Axud1E1 is regulated by neural plate border genes and multipotency factor SP5.

(A) Schematic representation of Axud1E1-300 mutants. Blue vertical bars represent MSX1, ZIC1 or SP5 binding motifs (see Fig 4M) that were disrupted in each construct. (B-D) Electroporation of the Axud1E1-300 mutants. Embryos were transfected with the wild-type enhancer on the left side and a mutant construct on the right side. Reduction of enhancer activity was observed when MSX1 (arrow in B) and ZIC1 (arrow in C) binding motifs were mutated. In contrast, mutation of SP5 motifs (D) resulted in an increase of Axud1E1-300 activity (upward arrow). (E) Quantification of the effect of MSX1 (n = 11), ZIC1 (n = 10) and SP5 (N = 10) binding site mutation on Axud1E1-300 activity. Statistical significance determined via ANOVA. (F-H) Effect of knockdown of putative regulators MSX1 (F), ZIC1 (G) and SP5 (H) on the activity of Axud1E1-300. MSX1 and ZIC1 knockdown result in reduction of Axud1E1-300 activity. SP5 morphant embryos display an increase in reporter expression indicating that SP5 represses Axud1E1. (I) Quantification of the effect of MSX1 (n = 12), ZIC1 (n = 12) and SP5 (N = 16) loss-of-function on Axud1E1-300. Statistical significance determined via ANOVA. (J-K) SP5 loss-of-function results in premature activation of Axud1E1 and AXUD1. qPCR for Axud1E1 (J) and AXUD1 (K) transcripts (HH6) in control neural plate border tissue vs neural plate border tissue transfected with SP5 morpholino. Error bars represent ± SEM. The statistical significance was determined via an unpaired t-test. (L) qPCR for AXUD1, ZIC3 and SALL4 transcripts in control vs neural folds transfected with an SP5 overexpression construct. Error bars represent ± SEM. The statistical significance was determined via an unpaired t-test. (M) In situ hybridization for AXUD1 after overexpression of SP5 within the right side of the embryo depicts reduction in AXUD1 expression levels (arrows). HH, Hamburger and Hamilton; mut, mutant; MO, morpholino; WT, wild type; bp, base pair; Scale bars represent 100μm (B-D, F-H), 500μm (M); *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Fig 6. SP5 interacts with enhancers of multiple neural crest genes.

(A-B) SP5 is transiently expressed in the neural crest lineage. Whole mount in situ hybridization shows expression of SP5 in the neural plate border (arrows in A). During neural crest specification, SP5 is excluded from the midline of the embryo (B). (C-D) Whole mount in situ hybridization shows that SP5 and AXUD1 are expressed in complementary expression patterns during neurula stages (HH9). AXUD1 transcripts are detected in the pre-migratory neural crest cells (D). (E) Diagram of CUT&RUN experiment used to map genome occupancy of SP5. (F) Genomic locations of SP5 binding events. The majority of SP5-associated regions are intergenic. (G) Heatmaps displaying SP5 and H3K27Ac signal at SP5-bound regions. (H) CUT&RUN profiles showing binding of SP5 and normal rabbit IgG at SP5-bound peaks. (I) CUT&RUN profiles of H3K27Ac, TFAP2A and SP5 at the Axud1E1-500 region. (J) Significantly enriched GO terms from Gene Ontology analysis of genes associated with SP5-bound regions. The diagram includes GO terms with Gene Ratio > 0.1. (K) CUT&RUN profiles of H3K27Ac, TFAP2A and SP5 at the enhancer Sox9E1 and the SP5 promoter. (L) Sub-circuit controlling the specification program in gastrula- and neurula-stage chicken embryos. SP5 and AXUD1 are activated by Wnt signaling. Sequential activation of NPB genes and the direct repression of Axud1E1 by SP5 restrict the timing of AXUD1 expression to specification stages. NPB, neural plate border; Kb, kilobase; HH, Hamburger and Hamilton; Scale bars represent 500μm (A-B), 200μm (C-D).