miRNA regulation of Sdf1 chemokine signaling provides genetic robustness to germ cell migration

Abstract

microRNAs (miRNAs) function as genetic rheostats to control gene output. Based on their role as modulators, it has been postulated that miRNAs canalize development and provide genetic robustness. Here, we uncover a previously unidentified regulatory layer of chemokine signaling by miRNAs that confers genetic robustness on primordial germ cell (PGC) migration. In zebrafish, PGCs are guided to the gonad by the ligand Sdf1a, which is regulated by the sequestration receptor Cxcr7b. We find that miR-430 regulates sdf1a and cxcr7 mRNAs. Using target protectors, we demonstrate that miR-430-mediated regulation of endogenous sdf1a (also known as cxcl12a) and cxcr7b (i) facilitates dynamic expression of sdf1a by clearing its mRNA from previous expression domains, (ii) modulates the levels of the decoy receptor Cxcr7b to avoid excessive depletion of Sdf1a and (iii) buffers against variation in gene dosage of chemokine signaling components to ensure accurate PGC migration. Our results indicate that losing miRNA-mediated regulation can expose otherwise buffered genetic lesions leading to developmental defects.

Figure 1

miR-430 target validation of chemokine signaling genes. (a) Schematic representation of the experimental set up. Expression of GFP reporters with the 3′UTRs of putative targets is compared in wild type and MZdicer embryos. dsRed mRNA lacking a target site was co-injected as a control. To test whether the miR-430 target site plays a role in the regulation, wild type embryos are also injected with wild type reporters (wt-3′UTR) or mutant reporters where three bases in the miR-430 target site were mutated (mut-3′UTR). All three miR-430 target sites in the sdf1a 3′UTR were mutated to generate the mutant reporter. (b) Quantification of the GFP fluorescence in MZdicer compared to wild type embryos injected with each GFP reporter mRNA and dsRed control. GFP fluorescence is normalized to the dsRed control. Data are shown as mean ± s.d. (c, d) Fluorescence microscopy shows GFP (green) and dsRed (red) expression in 24–28 hpf embryos injected with GFP reporters with sdf1a or cxcr7b 3′UTR. Endogenous miR-430 represses the expression of each reporter in wild type but not MZdicer embryos. Similarly, wild type embryos injected with the mut-3′UTR reporter fail to repress GFP. The sequence of the wild type or the mutant target site and miR-430 are shown in Supplementary Fig. 2.

Figure 2

Target Protectors prevent miRNA-mediated repression of target GFP reporters. (a) Schematic shows repression of targets by a miRNA and loss of repression of mutated reporters. GFP reporters are protected by TPs (purple) while binding of the control target protector (black) downstream of the miR-430 target site does not prevent miRNA-mediated repression. (b, c) GFP and dsRed fluorescence in 24-28 hpf embryos injected with GFP reporters. Expression of a wild type GFP-sdf1a 3′UTR reporter is repressed. Injection of sdf1a-TP, but not a control TP, blocks miR-430-mediated repression of the GFP reporter. Expression of the mut-3′UTR GFP reporter with the first (mut1-3′UTR) or all three target sites mutated (mut123-3′UTR) is shown for comparison. (c) Derepression of the GFP-cxcr7b wt-3′UTR reporter is also observed upon injection of the cxcr7b-TP. Predicted Watson-Crick pairing of the 3′UTR target sites with each TP (blue) and miR-430 are shown in Supplementary Fig. 3.

Figure 3

Blocking miR-430-mediated repression of sdf1a and cxcr7b causes PGC mislocalization and expanded sdf1a expression. (a) Schematic representation of the experimental setup. Injecting the TP (purple) blocks miRNA-mediated repression, increasing mRNA expression and leading to mislocalization of cells. Co-injecting a morpholino to reduce translation of the target gene (red, AUG MO) rescues the mislocalization phenotype. The inset shows the region of the embryo depicted in the panels (b, d-h). (b, d-g) Whole mount in situ of nanos mRNA, labeling PGCs in 24 hpf embryos. Bracket shows correct localization of PGCs. Arrowheads identify mislocalized PGCs. (c) Quantification of the percentage of embryos with mislocalized PGCs in each experimental condition as indicated. A significantly increased number of TP-injected embryos have mislocalized PGCs (*, p=1.185 ×10⁻⁷, sdf1a-TP; p=2.52×10⁻⁷, cxcr7b-TP; two-sided Fisher’s exact test). Error bars show ± s.d. (d, e) Representative images of PGC mislocalization are shown. (f, g) Co-injection of a low level of the corresponding AUG MO rescues the TP phenotype (sdf1a AUG MO, 0.01 pmol; cxcr7b AUG MO, 0.045 pmol). (h) In situ hybridization to detect sdf1a mRNA. The trunk of embryos at 20 hpf, 22 hpf, and 24 hpf are wild type, injected with sdf1a-TP, or MZdicer. Brackets illustrate the extension of the sdf1a expression domain along the pronephric region. (i) qPCR for sdf1a in 24 hpf wild type, sdf1a-TP-injected embryos, and MZdicer. An increase in expression was observed in the absence of miR-430-mediated repression. (j) Schematic summary of Sdf1a tail expression and the resulting PGC mislocalization.

Figure 4

miR-430 and Cxcr7b act in a functionally redundant manner to refine Sdf1a expression. (a, d) Quantification of PGC mislocalization. Embryos were injected at the one-cell stage with a morpholino targeting the start site (AUG MO) of cxcr7b (a) or sdf1a (d). These AUG MOs were injected at low concentrations, which were insufficient to completely knockdown the transcript and caused a weak mislocalization phenotype. (a) Co-injecting sdf1a-TP and cxcr7b AUG MO causes significantly more mismigration that injection of sdf1a-TP or the same amount of the AUG MO alone (*, p=4.08×10⁻³, sdf1a-TP + 45 fmol to 45 fmol alone; p=4.87×10⁻³, sdf1a-TP + 90 fmol to 90 fmol alone; two-tailed Fisher’s exact test), suggesting that miR-430 regulation of sdf1a mRNA can partially compensate for a reduction of cxcr7b. Similarly, co-injecting cxcr7b-TP and sdf1a AUG MO significantly enhances the mislocalization phenotype (*, p=8.93×10⁻⁴, cxcr7b-TP + 10 fmol to 10 fmol alone; p=0.016, cxcr7b-TP + 10 fmol to 10 fmol alone; two-tailed Fisher’s exact test), suggesting that regulation of cxcr7b by miR-430 prevents excessive clearance of the sdf1a. Data are shown as mean ± S.D. (b, e) nanos in situ at 24 hpf to visualize the location of germ cells. Brackets indicate correctly localized PGCs, and arrowheads show mislocalized cells. (c, f) Scheme representing the predicted effect of the experimental conditions on Sdf1a and Cxcr7b shown in panels (b and e). The added effect of removing miR-430 targeting and modulation by Cxcr7b supports a functional redundancy of miR-430 and Cxcr7b.

Figure 5

miR-430 buffers against overexpression of the chemokine signaling components. (a) Schematic representation of the experiment. mRNA encoding the open reading frame for the target gene with either the wt-3′UTR or a mutant 3′UTR with the miR-430 target site mutated is injected at the one-cell stage. PGC localization is assayed at 24 hpf using an in situ for nanos. The rectangle illustrates the region of the embryos shown in panels d, e and g. (b, c) Quantification of the percentage of embryos with PGCs outside of the gonad region upon injection of sdf1a mRNA (b) or cxcr7b mRNA (c). A significant difference is seen between transcripts with the wt-3′UTR (blue) and those with the mut-3′UTR (red) for sdf1a (*, p=7×10⁻⁶) and cxcr7b (*, p=6.3×10⁻³; two-tailed Fisher’s exact test). (d, e) Representative images of injected embryos are shown. The bracket illustrates the PGCs that are correctly localized. Arrowheads indicate mislocalized PGCs. Error bars indicate ± s.d.

Figure 6

Regulation by miR-430 guards against variation in gene dosage. (a) Quantification of the percentage of embryos with mislocalized PGCs in different experimental conditions as shown. Co-injection of sdf1a-TP, cxcr7b-TP, and a low level of translation blocking morpholino (MO) against cxcr7b (0.02 pmol), sdf1a (0.005 pmol), or cxcr4b (0.005 pmol) causes a significant increase in PGC mismigration (*, p=8.89×10⁻³, cxcr7b MO; p=2.87×10⁻³, sdf1a MO; p=0.011, cxcr4b MO; two-tailed Fisher’s exact test). A similar effect is seen upon injection of sdf1a-TP and cxcr7b-TP in cxcr4b heterozygous mutants (*, p=7.50×10⁻⁶) and cxcr7b heterozygous mutants (*, p=1.64×10⁻⁶) Error bars show ± s.d. (b) Representative examples of PGC mislocalization, shown by nanos in situ. Brackets show correctly localized PGCs, and arrowheads indicate mislocalized cells.

Figure 7

Model of miR-430-mediated repression of chemokine signaling. (a) Our results are consistent with a model in which miR-430 regulates sdf1a at the RNA level while previous results indicate that Cxcr7b, a decoy receptor, restricts the spatial expression pattern of Sdf1a protein. (b) Model adapted from to represent how miR- 430 regulates the dynamic expression of Sdf1a (blue gradient). miR-430-mediated regulation of sdf1a facilitates the formation of a sharp Sdf1a gradient by accelerating the clearance of sdf1a mRNA, concentrating expression on the actively transcribing domains (gray box) (middle panel). miR-430 modulates the levels of cxcr7b to prevent excessive clearance of the Sdf1a protein (right panel). (c) Model for generating robustness by regulating translation of abundantly transcribed genes. High levels of transcription coupled with inefficient translation can lower intrinsic noise in protein output,. (e) By dampening the expression of chemokine signaling genes, miR-430 buffers against changes in gene dosage (blue). We postulate that injecting TPs to block miR-430-mediated repression of sdf1a and cxcr7b increases the variability of gene expression (red). This reduces the ability of the system to compensate for minor perturbations in expression (See also Fig. 4, 5, and Supplementary Fig. 9)