An Epidermal MicroRNA Regulates Neuronal Migration Through Control of the Cellular Glycosylation State

See allHide authors and affiliations

Science  20 Sep 2013:
Vol. 341, Issue 6152, pp. 1404-1408
DOI: 10.1126/science.1242528

Extracellular Regulation

During Caenorhabditis elegans development, the hermaphrodite-specific neurons (HSNs) migrate and then extend axons toward their functional targets. Posttranslational modification of heparan sulfate proteoglycans are important for HSN development, and so Pedersen et al. (p. 1404) tested the effect of disrupting or reducing chondroitin and heparan sulfate synthesis during C. elegans development. The results suggest that proteoglycan biosynthesis is tightly regulated by a microRNA pathway to shape the cell surface glycosylation architecture required to direct neuronal migration.


An appropriate balance in glycosylation of proteoglycans is crucial for their ability to regulate animal development. Here, we report that the Caenorhabditis elegans microRNA mir-79, an ortholog of mammalian miR-9, controls sugar-chain homeostasis by targeting two proteins in the proteoglycan biosynthetic pathway: a chondroitin synthase (SQV-5; squashed vulva-5) and a uridine 5′-diphosphate–sugar transporter (SQV-7). Loss of mir-79 causes neurodevelopmental defects through SQV-5 and SQV-7 dysregulation in the epidermis. This results in a partial shutdown of heparan sulfate biosynthesis that impinges on a LON-2/glypican pathway and disrupts neuronal migration. Our results identify a regulatory axis controlled by a conserved microRNA that maintains proteoglycan homeostasis in cells.

Animal development requires the differentiation and assembly of distinct cell types into specific tissues and organs. During these processes, cells are guided by interactions with the extracellular matrix (ECM), which contains a variety of signaling molecules, including proteoglycans (1). Proteoglycans are made up of core proteins, such as syndecans, glypicans, and perlecans, that are decorated with varying numbers of long, unbranched glycosaminoglycan (GAG) chains. GAG chains vary in type and length and may be modified by sulfation and epimerization. Differential glycosylation and modification produce diverse interfaces for ligand-receptor interactions (2, 3). These structural parameters must be regulated to permit the coordination of specific and context-dependent intercellular signaling events (4, 5). The biosynthetic pathways that assemble and modify GAG chains on core proteins are highly conserved, and their disruption can cause developmental defects and lead to disease in many systems (4, 6, 7). GAG biosynthesis requires the transport of nucleotide sugars [uridine 5′-diphosphate (UDP)–sugars] from the cytoplasm to the Golgi apparatus (fig. S1). Generation of GAG chains commences with the addition of four specific UDP-sugars, which are added to defined attachment sites on core proteins to form a common core tetrasaccharide linkage region (fig. S1) (8). At this point, there is a bifurcation in the biosynthetic pathway where specific synthases and glycosyltransferases extend either chondroitin or heparan sulfate chains onto the core tetrasaccharide intermediate (fig. S1) (8). The supply of UDP-sugars and the relative expression levels of the respective biosynthetic enzymes inform the balance of heparan sulfate and chondroitin substitution, as exemplified by defects caused by disruption of the respective biosynthetic enzymes or UDP-sugar transporters (6, 7, 911).

We used Caenorhabditis elegans to study the biological requirements of proteoglycan homeostasis for neuronal development. The hermaphrodite-specific neurons (HSNs) migrate a long distance during embryogenesis. During larval development, the HSN axons navigate a precisely defined path (12, 13). These complex neurodevelopmental decisions are regulated by multiple conserved molecular pathways and the external environment (1215). As posttranslational modification of heparan sulfate proteoglycans is important for HSN development (14), we used this paradigm to test the effect of disrupting UDP-sugar supply or reducing chondroitin and heparan sulfate synthesis (Fig. 1). We tested the effect of disrupting proteoglycan homeostasis on HSN development using hypomorphic alleles of sqv-7 (squashed vulva), a UDP-sugar transporter required for GAG biosynthesis and the chondroitin synthase sqv-5 (Fig. 1, A and B) (1618). In addition, we performed RNA-mediated interference (RNAi) to knock down the heparan sulfate glycosyltransferases rib-1 and rib-2 (related to mammalian RIB-affecting EXT gene family) (Fig. 1, A and B). We analyzed HSN development using a transgenic strain in which the HSN cell bodies and axons are marked with green fluorescent protein (gfp). We found that similar defects in HSN development could be caused by any of three avenues: (i) reduction of UDP-sugar supply, (ii) disruption of chondroitin synthesis, or (iii) reduction of heparan sulfate synthesis (Fig. 1, A and B). These data suggest that regulatory mechanisms exist to maintain the correct balance of UDP-sugars supplied to the Golgi and to maintain the correct amounts of chondroitin and heparan sulfate biosynthetic enzymes.

Fig. 1 Proteoglycan homeostasis is essential for HSN development.

(A and B) Reduced function of sqv-7, sqv-5, rib-1, or rib-2 causes comparable HSN developmental defects. In wild-type animals (top panel), HSN cell bodies migrate to a position just posterior to the vulva and extend axons in a highly stereotypical manner. Their axons extend ventrally and then around the vulva before entering the ventral nerve cord, where each axon is separated by the hypodermal ridge. Hypomorphic alleles of sqv-5(k175) and sqv-7(n2839) (middle panels) as well as RNAi against rib-1 or rib-2 (bottom panels) cause similar HSN developmental defects. HSN cell bodies do not migrate to their correct position, and their axons are misguided. Vulval position is marked with a red asterisk, cell bodies with white arrowheads, and misguided axons by red arrowheads. Ventral view, anterior to the left. Scale bar: 20 μm. n > 50, *P < 0.05. Error bars represent means ± SEM.

We hypothesized that the regulatory mechanisms required to maintain proteoglycan homeostasis would need to be rapid and reversible owing to the demands of the changing molecular and cellular landscape during development. Thus, microRNAs (miRNAs) would be excellent candidate regulators of the proteoglycan biosynthetic pathway. We analyzed the 3′–untranslated regions (3′UTRs) of the sqv-5, sqv-7, rib-1, and rib-2 genes and found that only two conserved miRNA families are predicted to target these genes—the mir-79 and mir-124 families (table S1). We therefore examined knockout alleles of mir-75, mir-79, and mir-124 for defects in HSN development. We found that only loss of mir-79 caused defects in HSN cell migration and axon guidance (Fig. 2 and fig. S2A). The defects observed were similar to those caused by disruption of proteoglycan homeostasis. mir-79 is predicted to target both sqv-5 and sqv-7, suggesting that the mir-79 HSN phenotype is caused by dysregulation of these proteoglycan biosynthetic pathway genes.

Fig. 2 Loss of mir-79 phenocopies proteoglycan homeostasis-deficient HSN defects.

(A) HSN anatomy of mir-79(n4126) mutants. In mir-79(n4126) mutant animals, HSN cell bodies do not migrate to their correct position, and their axons are misguided in a pattern similar to that of sqv-5 and sqv-7 hypomorphic alleles and rib-1 and rib-2 knockdown (Fig. 1). Vulval position is marked with a red asterisk, cell bodies by white arrowheads, and misguided axons by red arrowheads. Ventral view, anterior to the left. Scale bar: 20 μm. (B) A mir-79prom::2xNLS::yfp transcriptional reporter drives expression in the hypodermis from mid-embryogenesis through to adult (see fig. S2D for larval pictures). Expression is first observed at the bean stage in hypodermal P cells (ventral) and in cells undergoing dorsal intercalation in addition to the lateral hypodermis (dorsal). At the comma and 2-fold stages, expression is observed in ventral, lateral, and dorsal hypodermis. Region used to drive yfp expression is shown in yellow on the genomic view (top). Upper panels are Nomarski micrographs and bottom panels are fluorescence images of the same embryos. Anterior to the left. Scale bar: 10 μm. (C) mir-75(4472) mutant animals do not exhibit defects in HSN development, and the mir-75(n4472); mir-79(n4126) double knockout does not show an additive effect when compared to the mir-79(n4126) mutant. Driving mir-79 expression using either its own or a heterologous hypodermal promoter (dpy-7) rescues mir-79(n4126)-induced HSN developmental defects. Driving mir-79 expression using a heterologous muscle promoter (unc-120) does not rescue mir-79(n4126)–induced HSN developmental defects. Driving expression of an unrelated miRNA (lsy-6) in the hypodermis does not rescue the mir-79(n4126)–induced HSN developmental defects. n > 50, **P < 0.01, ***P < 0.001; n.s., not significant. # refers to independent transgenic lines. Error bars represent means ± SEM.

mir-79 is highly expressed during embryogenesis and the L2/L3 larval stages (19) (fig. S2). The temporal pattern of mir-79 expression coincides with the developmental timing of the HSN migration and axon guidance events (12). To identify the tissue(s) in which mir-79 functions to regulate HSN development, we monitored the spatial and temporal expression pattern of mir-79 using the full intergenic sequence, upstream of the mir-79 hairpin, to drive nuclear-localized expression of the gene encoding yellow fluorescent protein (yfp). We found that expression of the mir-79prom::2xNLS::yfp reporter was first detected in embryonic hypodermal (epidermal) tissue at 330 min after fertilization and continued to be expressed in the hypodermis through larval development (Fig. 2B and fig. S2D).

Although vertebrate miR-9 family members are predominantly expressed in the nervous system (20), we did not detect expression of the mir-79 reporter in the C. elegans nervous system. Nonetheless, we found that expression of the mir-79 hairpin under the control of either its own promoter or a heterologous hypodermal-specific promoter partially rescued HSN migration defects of mir-79 mutant animals (Fig. 2C). mir-79 expression in muscle or expression of an unrelated miRNA (lsy-6) in the hypodermis could not rescue HSN migration defects (Fig. 2C). We did not observe any defects in hypodermal patterning or specification in the mir-79 mutant (fig. S3), suggesting that the HSN phenotypes observed are likely caused by a signaling rather than a structural defect. Taken together, these results indicate that mir-79 functions in the hypodermis to control HSN development.

Because the majority of miRNAs are known to inhibit translation, we hypothesized that the putative mir-79 targets, sqv-5 and sqv-7, would be up-regulated in the mir-79 mutant background, possibly in the hypodermis. We therefore examined whether the abundance of sqv-5 and sqv-7 mRNAs was increased in mir-79 mutant animals using quantitative polymerase chain reaction analysis of total RNAs from embryos (Fig. 3A). We found that sqv-5 and sqv-7 mRNA abundance is up-regulated in mir-79 mutant animals by 1.6-fold and 1.3-fold, respectively (Fig. 3A). These modest changes in mRNA abundance may indicate that up-regulation in the hypodermis is not easily detectable in the context of the whole organism or that mir-79 does not act to degrade these target mRNAs. To study the effect of mir-79 on SQV-5 and SQV-7 proteins, we constructed translational gfp reporter strains that express fluorescent fusion proteins upstream of the respective endogenous 3′UTRs. These integrated expression lines exhibit weak expression in wild-type embryos (Fig. 3C). When these same reporters were crossed into the mir-79 mutant, we observed that the expression of both SQV-5 and SQV-7 fusion proteins was increased at three stages of embryonic development that we examined (bean, 1.5-fold, and 3-fold stages) (Fig. 3C and fig. S4). These embryonic stages coincide with the time at which HSN neuronal migration occurs (12). These data suggest that overexpression of sqv-5 and sqv-7 in mir-79 mutant animals causes HSN developmental defects.

Fig. 3 mir-79–induced HSN defects are caused by dysregulated sqv-5 and sqv-7.

(A) sqv-5 and sqv-7 mRNA is up-regulated in mir-79–null embryos. Measurements were taken in triplicate. (B) Knockdown of sqv-5 or sqv-7 using RNAi fully suppresses mir-79(n4126) HSN developmental defects. n > 100, ***P < 0.001. (C) Expression of sqv-5 and sqv-7 translational gfp reporter transgenes under the control of their respective endogenous 3′UTRs. mir-79(n4126) mutant animals exhibit increased expression of the fluorescent fusion proteins at three stages of embryonic development. In fluorescent images of bean and 1.5-fold–stage embryos, the field of HSN migration is marked (white dashed line). Scoring of fluorescence intensity of 1.5-fold embryos is detailed in fig. S4. Upper panels are Nomarski micrographs and bottom panels are fluorescence images of the same embryos. Anterior to the left. Scale bar: 10 μm. (D and E) Overexpression of sqv-5 or sqv-7 in the hypodermis phenocopies the HSN developmental defects of mir-79(n4126) mutant animals. The penetrance of HSN developmental defects were similar to those observed with loss of mir-79, as were the severity of phenotypes. Vulval position is marked with a red asterisk, cell bodies with white arrowheads, and misguided axons by red arrowheads. Ventral view, anterior to the left. Scale bar: 20 μm. n > 50, *P < 0.05, **P < 0.01, ***P < 0.001. # refers to independent transgenic lines. All error bars represent means ± SEM.

We used two approaches to confirm whether overexpression of sqv-5 and sqv-7 causes the HSN developmental defects in the mir-79 mutant. First, we used RNAi to knock down the expression of sqv-5 and sqv-7 in the mir-79 mutant and evaluated suppression of the HSN migration phenotype (Fig. 3B). RNAi knockdown of either sqv-5 or sqv-7 fully suppressed the mir-79 HSN migration phenotype to background levels (Fig. 3B). As the mir-79 HSN migration defect is fully suppressed by RNAi knockdown of either sqv-5 or sqv-7, we next hypothesized that overexpression of these proteins in the hypodermis (the site of mir-79 action) would phenocopy the mir-79 phenotype in wild-type animals. We found that hypodermal overexpression of sqv-5 or sqv-7 causes quantitatively and qualitatively HSN migration defects similar to those of the mir-79 mutant (Fig. 3, D and E). Taken together, these data suggest that the correct balance of expression of GAG biosynthesis pathway genes is required for HSN development in C. elegans.

To assess the possible direct regulation of sqv-5 and sqv-7 expression by mir-79 in vivo, we used a heterologous 3′UTR sensor assay, which had been previously used to measure the activity of the let-7 miRNA (21). We generated one transgene on which we expressed two reporters in the pharynx: one green fluorescent protein (gfp) “sensor” reporter with the 3′UTR of either sqv-5 or sqv-7 and another with a red fluorescent protein (mCherry) reporter under the control of the unc-54 3′UTR, which does not contain any mir-79 binding sites. On a second transgene, we also expressed mir-79 in the pharynx. As expected, we found that expression of mir-79 robustly down-regulated gfp expression (controlled through sqv-5 or sqv-7 3′UTR sequence) but not mCherry expression (fig. S5). These data suggest that mir-79 can directly interact with the sqv-5 and sqv-7 3′UTRs to down-regulate their expression. mir-79 regulation of the sqv-5 and sqv-7 3′UTRs was lost when the mir-79 target sites were mutated (fig. S5). We found that the Drosophila and human orthologs of SQV-5—CG9220 and CHSY1, respectively—are also predicted targets of miR-9 (fig. S6) and, in the case of CHSY1, this relationship is also suggested by PAR-CLIP data (22).

To analyze the impact of imbalance in the proteoglycan biosynthetic pathway on the glycosylation of core proteins, we isolated intact proteoglycans from wild-type and mir-79 mutant animals (Fig. 4, A to C). Proteoglycans were treated with heparinase III or chondroitinase ABC, which cleave heparan sulfate and chondroitin chains, respectively. Removal of the chains result in “stub” oligosaccharides, which are recognized by antibodies against heparan sulfate and chondroitin (23, 24). Western blot analysis using heparan sulfate– and chondroitin-specific antibodies showed that in mir-79 mutant animals, the expression level of heparan sulfated proteoglycans is reduced by a factor of 10, whereas the level of proteoglycans with chondroitin chains is unaffected (Fig. 4, A to C). Overexpression of rib-2, one of the two glycosyltransferases required for heparan sulfate synthesis in C. elegans (25), provided sufficient heparan sulfate to rescue mir-79–induced HSN defects (Fig. 4D). Taken together, our biochemical and genetic analyses suggest that the neurodevelopmental defects observed in the mir-79 mutant are caused by defects in the biosynthesis of heparan sulfate proteoglycan in the hypodermis.

Fig. 4 Heparan-sulfated LON-2/glypican acts downstream of mir-79 to direct neuronal guidance.

(A to C) Intact proteoglycans were isolated from N2 or mir-79(n4126) embryos and treated with heparinase III or chondroitinase ABC. Western blot analysis was performed using monoclonal antibodies specific for heparan sulfate (A) or chondroitin (B). Antibodies only detect heparan sulfate or chondroitin stubs that are revealed after enzymatic digestion. Actin was used to normalize expression. Three Western blots from independent samples were performed, the mean values of which are depicted in (C), and representative examples are shown in (A) and (B). Data are presented as wild-type set to one-fold. Error bars represent SE. **P < 0.01. (D) Driving rib-2 expression under the control of its own promoter rescues mir-79(n4126)–induced HSN developmental defects. n > 50, ***P < 0.001. # refers to independent transgenic lines. (E) HSN anatomy of lon-2(e678), mir-79(n4126); lon-2(e678) and lon-2(e678); dbl-1(wk70) mutant animals. Vulval position is marked with a red asterisk, cell bodies by white arrowheads, and misguided axons by red arrowheads. Ventral view, anterior to the left. Scale bar: 20 μm. (F) Scoring of HSN developmental defects in mir-79(n4126), lon-2(e678), mir-79(n4126); lon-2(e678), dbl-1(wk70), and lon-2(e678); dbl-1(wk70) mutant animals. n > 50. The mir-79(n4126); lon-2(e678) double-mutant HSN phenotype is not significantly different from either single mutant. In addition, the lon-2(e678); dbl-1(wk70) double-mutant HSN phenotype is not significantly different from the lon-2(e678) single mutant. n > 70. Rescue scoring of HSN developmental defects in lon-2(e678) mutant animals shows that wild-type LON-2 protein (lon-2prom::LON-2) fully rescues the mutant phenotype, whereas mutant LON-2 protein, with the mutated GAG attachment sites (lon-2prom::LON-2ΔGAG sites), is unable to rescue. n > 30. The mutated LON-2 protein does rescue the Lon phenotype (data not shown). *P < 0.05; n.s., not significant. All error bars represent means ± SEM. (G) mir-79 is expressed in the hypodermis, where it regulates the expression of two proteoglycan biosynthesis pathway genes (sqv-5 and sqv-7). Correct control of this pathway during embryogenesis enables faithful heparan sulfate substitution of LON-2/glypican, which directs HSN neuronal development through an as-yet unknown mechanism.

There are four canonical heparan sulfate proteoglycan core proteins in the C. elegans genome: lon-2/glypican, gpn-1/glypican, sdn-1/syndecan, and unc-52/perlecan (2). We focused on lon-2 and its role in HSN migration, as it is expressed in the hypodermis (26), where mir-79 functions, and lon-2–null mutant animals exhibit HSN phenotypes similar to those of mir-79 mutant animals (Fig. 4E). We asked whether the mir-79 HSN defects are caused by defective LON-2 signaling by performing genetic double-mutant analysis. Null mutations of mir-79 and lon-2 exhibit similar penetrance of HSN developmental defects (Fig. 4F). If mir-79 functions in a pathway parallel to that of lon-2 rather in the same pathway, we would expect the phenotype of the double mutant to be additive. However, we found that the penetrance of HSN defects of mir-79; lon-2 double-mutant animals was not significantly different from either single mutant, suggesting that mir-79 and lon-2 act in the same pathway to regulate HSN development (Fig. 4F). The three GAG attachment sites on the LON-2 core protein are unnecessary for LON-2 regulation of body length (27). Our work would imply that these GAG attachment sites are crucial for the regulation of HSN migration owing to the requirement of LON-2 heparan sulfate chains. Indeed, we found that LON-2 cannot direct HSN development without intact GAG attachment sites (Fig. 4F). To regulate body length in C. elegans, lon-2 acts to repress the dbl-1/TGFβ (transforming growth factor–β) pathway through the core protein and independently of GAG attachment (26, 27). This suggests that the HSN developmental defects of lon-2 mutant animals are also independent of the dbl-1 pathway. Indeed, lon-2; dbl-1 double mutant animals exhibit the lon-2 mutant HSN phenotype, indicating that LON-2/glypican regulates HSN neuronal development through a pathway independent of the dbl-1/TGFβ pathway that governs body length (Fig. 4F).

We have identified a regulatory axis through which the highly conserved mir-79 gene controls a specific neurodevelopmental event (HSN cell migration and axon guidance) in C. elegans (Fig. 4G). mir-79 regulates this event non–cell-autonomously by controlling hypodermal expression of two components of the GAG biosynthetic pathway, sqv-5 and sqv-7, which encode a chondroitin synthase and UDP-sugar transporter, respectively. Aberrant expression of these proteins in the mir-79 deletion mutant reduces heparan sulfate biosynthesis and disrupts LON-2/glypican signaling. Our results imply that proteoglycan biosynthesis is tightly regulated to shape cell-surface glycosylation architecture required to direct neuronal migration—a control we have shown to be mediated by a microRNA.

Supplementary Materials

Materials and Methods

Figs. S1 to S6

Tables S1 and S2

References (2833)

References and Notes

  1. Acknowledgments: We thank O. Hobert, H. Buelow, P. Brodersen, and A. Lund for comments on the manuscript and T. Gumienny for lon-2 strains and reagents. Some strains used in this study were provided by the Caenorhabditis Genetics Center, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440). This work was supported by grants from the European Research Council (ERC Starting Grant no. 260807 to R.P.), the Lundbeck Foundation (Project nos. R118-A11481 to R.P. and R44-A4407 to J.R.C.), the Danish National Research Foundation, and the Novo Nordisk Foundation (to J.R.C.).
View Abstract

Stay Connected to Science

Navigate This Article