Research ArticleDevelopmental Biology

Sphingosine 1-phosphate stimulates eyelid closure in the developing rat by stimulating EGFR signaling

See allHide authors and affiliations

Science Signaling  23 Oct 2018:
Vol. 11, Issue 553, eaat1470
DOI: 10.1126/scisignal.aat1470

A signaling lipid directs eyelid morphogenesis

To prevent the eye from being damaged during birth and early postnatal development, the eyelids of many mammals migrate over the cornea and fuse during embryonic development. Failure of this process causes an eyes-open-at-birth (EOB) phenotype, which leads to permanent corneal damage. Bian et al. identified an insertional mutation in Spns2, which encodes a sphingosine 1-phosphate (S1P) transporter, in a strain of transgenic rats exhibiting the EOB phenotype. Experiments in vivo and with eyelid tissue explants in vitro showed that S1P stimulated eyelid closure by promoting the activation of proteins involved in cell migration and by stimulating epidermal growth factor receptor (EGFR) signaling. These results identify a complex network of signaling and feedback events through which S1P stimulates the metalloprotease-dependent mobilization of EGFR ligands to coordinate the migration and fusion of the eyelids during embryogenesis.

Abstract

In many mammals, the eyelids migrate over the eye and fuse during embryogenesis to protect the cornea from damage during birth and early life. Loss-of-function mutations affecting the epidermal growth factor receptor (EGFR) signaling pathway cause an eyes-open-at-birth (EOB) phenotype in rodents. We identified an insertional mutation in Spinster homolog 2 (Spns2) in a strain of transgenic rats exhibiting the EOB phenotype. Spns2, a sphingosine 1-phosphate (S1P) transporter that releases S1P from cells, was enriched at the tip of developing eyelids in wild-type rat embryos. Spns2 expression or treatment with S1P or any one of several EGFR ligands rescued the EOB Spns2 mutant phenotype in vivo and in tissue explants in vitro and rescued the formation of stress fibers in primary keratinocytes from mutants. S1P signaled through the receptors S1PR1, S1PR2, and S1PR3 to activate extracellular signal–regulated kinase (ERK) and EGFR-dependent mitogen-activated protein kinase kinase kinase 1 (MEKK1)–c-Jun signaling. S1P also induced the nuclear translocation of the transcription factor MAL in a manner dependent on EGFR signaling. MAL and c-Jun stimulated the expression of the microRNAs miR-21 and miR-222, both of which target the metalloprotease inhibitor TIMP3, thus promoting metalloprotease activity. The metalloproteases ADAM10 and ADAM17 stimulated EGFR signaling by cleaving a membrane-anchored form of EGF to release the ligand. Our results outline a network by which S1P transactivates EGFR signaling through a complex mechanism involving feedback between several intra- and extracellular molecules to promote eyelid fusion in the developing rat.

INTRODUCTION

The closing of the eyelids during fetal development is an evolutionarily conserved morphogenetic event requiring proliferation, differentiation, cytoskeleton reorganization, and migration of the epithelial cells at the tip of the developing eyelids (1). Mouse eyelid formation commences on embryonic day 13 (E13), with folds of surface ectoderm adjacent to the developing eye as the eyelid root. Eyelid closure begins at E14.5, when the leading edge of the root tip epithelium begins to extend toward the center of the eye, and is complete between E15.5 and E16.5 when the leading edges from the upper and lower lids meet and fuse, thus covering the entire ocular surface and forming a protective barrier over the cornea until the eyelids open about 12 days after birth (1). The cellular events involved in eyelid development are delayed by about 2 days in the rat compared with the mouse (1). Eyelid closure or fusion is thought to be mechanistically similar to that of the migrations of epithelial sheets in wound healing and neural tube closure (2). On the basis of genetic mutations that lead to an eyes-open-at-birth (EOB) phenotype in mice, it is clear that many complex signaling pathways regulate eye closure, including Wnt, sonic hedgehog, bone morphogenetic protein and activin, fibroblast growth factor receptor 2 (FGFR2), epidermal growth factor receptor (EGFR), G protein–coupled receptor 48 (GPCR48), and sphingosine 1-phosphate receptor 2 (S1PR2) and S1PR3 (S1PR2/3) cascades (311). In addition, eyelid closure requires several intracellular kinases, such as mitogen-activated protein kinase kinase kinase 1 (MEKK1), c-Jun N-terminal kinase (JNK), Rho-associated kinase (ROCK), and cadherin-1 (CDH1), as well as nuclear transcription factors, such as c-Jun, fos-related antigen–2 (Fra-2), forkhead box L2 (FOXL2), FOXC1, FOXC2, grainy head–like 3/LIM-only domain protein (Grhl3/Lmo4), and serum response factor (SRF) (8, 1226). Eyelid closure is associated with cell shape changes that are mediated by actin stress fibers and filopodia, key features of epithelial cell migration, at the leading edge of the sheet (27). A previous study reported that the MEKK1–c-Jun regulatory axis connects activin and transforming growth factor–β (TGF-β) signaling to the TGF-α–EGFR-ERK (extracellular signal–regulated kinase) or EGFR-RhoA-ROCK pathways to contribute to eyelid fusion (1719, 2830). Rats that exhibit an EOB phenotype due to insertion of a transgene encoding enhanced green fluorescent protein (EGFP) into a relevant gene are ideal models for exploring the mechanisms that control epithelial cell migration and epithelial sheet movement (27).

S1P, a potent bioactive sphingolipid metabolite, is involved in diverse cellular processes (31, 32). Cells metabolize the ubiquitous sphingolipid sphingomyelin to produce sphingosine, which can be either metabolized or phosphorylated by cytosolic sphingosine kinase 1 (SphK1) or nuclear SphK2 to yield S1P. Extracellular S1P stimulates five cell surface GPCRs, S1PR1 to S1PR5, that couple to distinct subsets of heterotrimeric G protein α subunits to activate signaling pathways, controlling diverse physiological functions (3335). S1P also functions as a second messenger that acts on intracellular targets (36, 37). Intracellular S1P acts as a stimulatory cofactor in promoting the E3 ubiquitin ligase activity of TNF receptor–associated factor 2, leading to the ubiquitination of receptor-interacting protein-1 and the activation of nuclear factor κB (37). Intracellular S1P can also act as an inhibitory cofactor; for example, nuclear S1P interacts with and inhibits the enzymatic activities of the histone deacetylases (HDACs) HDAC1 and HDAC2 (36). As with any signaling molecule, the abundance of S1P in cells is tightly controlled by its rapid degradation through two different mechanisms: S1P can be dephosphorylated by either lipid phosphatases or S1P-specific phosphatases (SPP1 and SPP2) to generate sphingosine or it can be irreversibly cleaved by S1P lyase as the final degradative step of sphingolipid metabolism (38).

S1P is synthesized intracellularly and released from cells by S1P transporters, after which it can activate its receptors (S1PR1 to S1PR5) in an autocrine or paracrine fashion, or both, a process known as S1P “inside-out” signaling (31, 35). Several adenosine 5′-triphosphate–binding cassette (ABC) transporters have been implicated in S1P export in vitro, such as ABCA1 in astrocytes (39), ABCC1 in mast cells and breast cancer cells (4043), ABCG2 in breast cancer cells (40), and other ABC family members in platelets and erythrocytes (40, 44, 45). However, the physiological role of ABC transporters in S1P export remains unclear because S1P export is unaffected in mice deficient in ABCA1, ABCA7, or ABCC1 (46). A member of the major facilitator superfamily of transporters, Spinster homolog 2 (Spns2), has been identified as an S1P transporter in zebrafish (47, 48). Spns2 is a putative 12–transmembrane domain protein belonging to the Spinster family (Spns1 to Spns3 in vertebrates and Spinster in Drosophila melanogaster) (49). The biological role of Spns2 was first elucidated from two zebrafish Spns2 mutants, two of hearts (toh) and ko157 (47, 48, 50). Both mutants developed a cardia bifida phenotype due to failed cardiac progenitor migration, which is similar to the phenotype observed in miles apart (mil, homologous to mammalian s1pr2) zebrafish mutants (47, 48, 51). Injection of Spns2 mRNA into mutant embryos or S1P into the yolk syncytial layer restores normal heart formation in mil mutants (48). Mil also regulates jaw development and cooperates with S1PR1 to control angiogenesis in zebrafish (52, 53). Spns2 deficiency in mammals causes plasma S1P abundance to decrease and impairs S1P-dependent lymphocyte trafficking with no apparent defects in heart development (5458). Spns2 exports S1P from mammalian cells, and increasing the concentration of plasma S1P is responsible for the egress of mature T and immature B cells from the thymus and bone marrow, respectively (5962). In addition to abnormalities in the immune system, loss of auditory and ocular functions, such as EOB and symblepharon, has also been observed in Spns2-deficient mice (54, 55, 58, 63). However, the mechanism of the Spns2-associated EOB phenotype has not been investigated.

In this study, we characterized a strain of transgenic green rat (GR) harboring an Egfp-marked transgene that displays the EOB phenotype (GREOB) with recessive inheritance and found Spns2 to be mutated by a transgene insertion in this strain. Transgenic expression of Spns2 or treatment with EGF or S1P rescued the EOB phenotype efficiently in vivo and in cultured eyelid explants in vitro. We found that Spns2-mediated export of S1P promoted keratinocyte migration by activating S1PR1, S1PR2, and S1PR3, which transactivated EGFR signaling by inducing the shedding of EGFR ligands mediated by the metalloproteases a disintegrin and metalloprotease 10 (ADAM10) and ADAM17. Keratinocytes from Spns2 mutants exhibited dysfunction of c-Jun and myocardin-related transcription factor A (MAL) that led to a decrease in the amount of the microRNAs (miRNAs) miR-21 and miR-222, both of which target the ADAM10 and ADAM17 inhibitor TIMP3 (tissue inhibitor of metalloproteases 3). Our results outline a complex cascade in which S1P is exported from eyelid epithelial cells by Spns2 and controls the closing of eyelids in a manner that depends on the activation of EGFR signaling in the leading edge of migrating eyelid epithelia in developing rats.

RESULTS

The failure of eyelid closure in the EGFP insertion line is due to disruption of Spns2 in GREOB

Some offspring of crosses between individuals from a strain of GR carrying a CAG-Egfp transgene displayed the EOB phenotype (Fig. 1, A and B). Because embryonic eyelid failure closure fails, these rats exhibited various abnormalities of the cornea, including opacity, ulceration, xerosis (dryness), and the absence of a corneal reflex (involuntary blinking in response to touch). These corneal phenotypes displayed an autosomal recessive pattern of inheritance (Fig. 1A, fig. S1A, and movie S1) and were linked to the expression of EGFP and distributed to offspring in a Mendelian fashion (Fig. 1B and fig. S1A). These results led us to conclude that EOB and corneal phenotypes were most likely the result of a single insertional mutation introduced by the Egfp transgene.

Fig. 1 Mapping the CAG-Egfp insertion site in the GR genome.

(A) Representative images showing the morphology of the cornea in wild-type (WT) rats, GR with no corneal phenotype (GR), and GR with corneal phenotype (GREOB) (n > 50 rats for each group). (B) Pedigree analysis of the corneal phenotype and EGFP expression in five generations. (C) CAG-Egfp (gray box) was inserted into exon 1 of Spns2, which has 13 exons. Pst I and Bgl II sites were identified near the CAG-Egfp insertion site to generate predicted products of 7227 and 3871 base pairs (bp), respectively. (D) Genomic DNA digested with Pst I or Bgl II was used for Southern blot analysis using a probe complementary to Egfp. A single band corresponding to the Pst I and Bgl II fragments (arrowheads) was identified in each genomic DNA sample from GRs with corneal phenotypes (GREOB) but not in genomic DNA from WT rats. (E) RT-PCR using primers P7 and P8 to detect Spns2 mRNA the brain and skin of WT and GREOB rats (n = 3 rats for each group).

Using vector sequences from the transgene for inverse polymerase chain reaction (PCR), we sequenced the genomic DNA, flanking the transgene to identify the position of the CAG-Egfp insertion in the genome (fig. S1, B to D). The transgene was inserted between positions 59,047,922 and 59,048,251 bp in NC_005109.4 of rat chromosome 10 (fig. S1, D and F to I), causing a 120-bp loss in the length of the genomic fragment (fig. S1, F to I). We screened the genes surrounding the insertion site and found that the transgene was located in intron 1 of Spns2 (Fig. 1C) (54). Southern blot analysis confirmed that only one copy of the CAG-Eg fp transgene was inserted into the GR genome (Fig. 1, C and D), suggesting that the phenotypes in this strain were due to disruption of Spns2. Last, whole-genome sequencing (WGS) of the GR strain confirmed that a single copy of the CAG-Egfp transgene was inserted into intron 1 of Spns2 in chromosome 10, which is consistent with the results from our inverse PCR sequencing and Southern blot analyses (Fig. 1D). These results indicate that one copy of exogenous Eg fp was inserted into intron 1 of Spns2 in chromosome 10, preventing the production of Spns2 mRNA. This RNA-null allele therefore represents an Spns2 deficiency and is hereafter referred to as Spns2ins/ins.

Using a forward primer spanning the boundary between exons 1 and 2 (P7) and a reverse primer from exon 4 (P8) for reverse transcription PCR (RT-PCR) (fig. S1E), we detected no Spns2 transcripts in extracts of brain and skin tissues from animals displaying the EOB and corneal phenotypes (Fig. 1E). To further verify the lack of expression of Spns2 in these rats, we assayed Spns2 transcripts by both quantitative PCR (qPCR) and in situ hybridization. By qPCR, using a new pair of primers (table S1; Spns2-1 and Spns2-2) to amplify the sequence between exon 1 and exon 3, we detected no Spns2 expression in the tissues of Spns2ins/ins rats (fig. S2A). In situ hybridization assays using a digoxigenin-conjugated RNA probe complementary to a sequence within exon 1 (primers P9 and P10; table S1) showed that Spns2 was mainly expressed in glomeruli in wild-type rats, whereas no positive signal was detected in Spns2ins/ins animals (fig. S2B).

Spns2 deficiency prevents eyelid closure in utero and causes an abnormal corneal phenotype

The corneas of adult Spns2ins/ins rats were opaque, with a thickened epithelium and inflammatory infiltration (Fig. 1A and fig. S4A). Furthermore, the EOB phenotype exhibited 100% penetrance in Spns2-deficient pups in contrast to wild-type rats, which are born with their eyes closed (Fig. 2A) (1). Embryonically closed eyelids serve as a protective barrier that is crucial for normal eye development. In the absence of eyelid fusion, the unprotected cornea is prone to damage after birth (1, 4). During development, migration of the leading edge of the eyelid root epithelium over the developing eye is critical for the closing of eyelids (1, 64). At E16.5, the eyes of all wild-type, Spns2ins/+, and Spns2ins/ins fetuses were open, with no apparent morphological differences at the eyelid margins (Fig. 2B). At E17.5, the leading edges of the rudimentary eyelid tips had begun to extend over the eye in wild-type and Spns2ins/+ fetuses but failed to extend in Spns2ins/ins fetuses (Fig. 2B). At E18.5, the two leading edges from both sides of the eyelid root fused at the center of the eye and covered the ocular surface in both wild-type and Spns2ins/+ rats, whereas the eye was still open without fusion of the upper and lower eyelids in Spns2ins/ins rats (Fig. 2B). Our results are consistent with a report by Hisano et al. (54) that Spns2-deficient mice display an EOB phenotype and that rat Spns2 shares 99% amino acid sequence identity with mouse Spns2 (fig. S3).

Fig. 2 EOB phenotype in Spns2ins/ins rats.

(A) Images illustrating the EOB phenotype in Spns2ins/ins rats at postnatal day 1 (P1) and P7. (B) Hematoxylin and eosin–stained eyes from WT, Spn2ins/+, and Spns2ins/ins rats at E16.5, E17.5, and E18.5. Arrows indicate the leading edges of the eyelids. Asterisks indicate the fusion site of the eyelids (n = 5 rats for each group). Scale bar, 250 μm. (C) In situ hybridization showing Spns2 mRNA in the leading edge of the eyelid of WT rats at E16.5 and E17.5 and at the eyelid fusion site at E18.5 (n = 3 rats for each group). Scale bar, 50 μm. (D) Immunofluorescence images showing the distribution of Spns2 protein in the eyelid of WT, Spn2ins/+, and Spns2ins/ins rats at E16.5, E17.5, and E18.5 (n = 5 rats for each group). Scale bar, 50 μm. (E) Eyes of Spns2ins/ins embryos expressing empty vector or Spns2. The boxed area in each panel is magnified in the inset. Arrows indicate the eyelid (n = 6 rats for each group). Scale bar, 500 μm.

By RNA in situ hybridization, Spns2 was expressed in the eyelid epidermis, but not in the eyelid conjunctiva, at E16.5, at the leading edge of the eyelid epidermis at E17.5, and at the site of eyelid fusion at E18.5 in wild-type rats (Fig. 2C). These results were confirmed by immunohistochemistry using a polyclonal antibody recognizing Spns2 that we generated and tested for specificity in HeLa cells expressing rat Spns2 (fig. S4E). Spns2 localized to the leading edge of the epidermis and the fusion site of eyelids at E17.5 and E18.5, respectively, in wild-type rats, but not to any eyelid cells in Spns2ins/ins rats (Fig. 2D). Lipofection or electroporation of a plasmid encoding Spns2 into Spns2ins/ins embryos at E16.5 (Fig. 2E) indicated that Spns2 is indispensable for eyelid closure in developing rats.

We also observed that lymphocyte, monocyte, neutrophil, eosinophil, and basophil counts were lower in adult Spns2ins/ins rats than in wild-type rats (fig. S4, B and C). The number and ratio of CD3+ T and B220+ B lymphocytes were reduced in the peripheral blood of Spns2ins/ins rats, consistent with lymphopenia in the mutant mice (fig. S4D) (55, 58). Postnatal retinal development was also abnormal, and we observed sporadic rosette structures posterior to the photoreceptor layers in Spns2ins/ins rats (fig. S4A). In contrast, the retinas of age-matched wild-type and Spns2ins/+ rats were normal, with no retinal infolding or rosette formation (fig. S4A). These results indicate that Spns2 has multiple roles in hematopoietic and eye development in rats.

S1P and EGF rescue abnormal stress fiber formation in keratinocytes and the EOB phenotype in Spns2-deficient rats

We investigated the abundance of S1P in the amniotic fluid of Spns2ins/ins rats during development. By enzyme-linked immunosorbent assay (ELISA), the abundance of S1P in amniotic fluid was significantly lower in Spns2ins/ins rats than in wild-type rats at E16.5 to E18.5 (Fig. 3A). To determine whether the phenotypes observed in mutant rats result from reduced extracellular S1P rather than from the intracellular accumulation of S1P, we cultured keratinocytes from wild-type and Spns2ins/ins neonatal rats and examined the abundance of S1P in both the culture supernatant and cell lysates by ELISA (fig. S5). The keratinocytes from wild-type rats secreted much more S1P into the medium than did keratinocytes from mutants, which is consistent with the measurements in amniotic fluid (fig. S5A). Both groups of keratinocytes showed similar amounts of intracellular S1P (fig. S5B).

Fig. 3 Keratinocyte migration is abnormal in Spns2ins/ins rats.

(A) S1P concentration in the amniotic fluid of WT and Spns2ins/ins rats (n = 4 rats for each group). (B) Immunofluorescence images showing phalloidin-positive stress fibers in keratinocytes (keratin 5 positive) from WT (+/+) and Spns2ins/ins (ins/ins) rats (n = 9 wells from three separate harvests). Scale bar, 25 μm. (C) Stress fibers in primary keratinocytes from Spns2ins/ins embryos treated with the indicated concentrations of S1P in vitro (n = 9 wells from three separate cell harvests). Scale bar, 25 μm. (D) Stress fibers in keratinocytes from Spns2ins/ins embryos treated with the indicated concentrations of EGF (n = 9 wells from three separate harvests). Scale bar, 25 μm. (E) HB-EGF and EGF concentrations in the medium of cultured keratinocytes from WT and Spns2ins/ins embryos (n = 4 rats for each group). (F) Embryonic eyelid tissue explants from Spns2ins/ins embryos that were untreated or treated with S1P or EGF. Arrows indicate the fused eyelids (n = 4 rats for each group). Scale bar, 200 μm. (G) Phosphate-buffered saline (PBS), methanol, FGF10, S1P, EGF, or HB-EGF was injected into the amniotic sac of Spns2ins/ins embryos at E16.5, and the eye phenotype was observed at E20.5 (n = 6 rats for each group). The boxed area in each panel is magnified in the inset. Arrows indicate the edge of the eyelid. Scale bar, 500 μm. Data in (A) to (E) are presented as means ± SEM. *P < 0.05, ***P < 0.001.

Embryonic eyelid closure is associated with cell shape changes and the formation of actin stress fibers in the developing eyelid epithelium, both of which are key features of epithelial cell migration (27). Phalloidin labeling of actin indicated that stress fibers were absent in keratinocytes from Spns2ins/ins rats compared with those from wild-type rats (Fig. 3B). We quantified the number of stress fibers within keratinocytes, and cells with three or more stress fibers were considered as positive for stress fiber formation. The results indicated that stress fiber formation was impaired in keratinocytes from Spns2ins/ins rats (fig. S6A), but application of S1P rescued this phenotype in a dose-dependent manner (Fig. 3C and fig. S6B). EGFR signaling is important for eyelid closure, because point mutation and knockout alleles of Eg fr or genes encoding EGFR ligands give rise to EOB phenotypes in mice (3, 4, 6, 65, 66). We found the EOB phenotype to be 100% penetrant in Spns2ins/ins rats, which is consistent with Egfr deficiency or the hypomorphic Egfr mutation wa-2 in mice (6, 66). Treatment of keratinocytes from Spns2ins/ins rats with various concentrations of EGF restored stress fiber formation (Fig. 3D and fig. S6C). The EGFR ligands EGF and heparin-binding EGF-like growth factor (HB-EGF) were less abundant in the culture medium of Spns2ins/ins keratinocytes than in wild-type cells (Fig. 3E).

Eyelid tissues explanted from Spns2ins/ins embryos did not undergo closure in culture, but treatment with S1P or EGF induced closure (Fig. 3F). Injecting S1P, EGF, or HB-EGF, but not fibroblast growth factor 10 (FGF10) or vehicle, into the amniotic sac of Spns2ins/ins embryos at E16.5 also induced eyelid closure in vivo (Fig. 3G). These results are consistent with eyelid closure, requiring the Spns2-mediated release of S1P into the extracellular environment to induce stress fiber formation and epithelial cell migration through a mechanism that depends on EGFR signaling.

Spns2 deficiency disrupts EGFR signaling in keratinocytes

There were no significant differences in the abundances of EGFR or ERK1 and ERK2 (ERK1/2) in eyelids from wild-type, Spns2ins/+, and Spns2ins/ins embryos at E17.5 and E18.5 (Fig. 4A and fig. S7, A and B). However, the amounts of EGFR phosphorylated at Tyr1068 and ERK1/2 phosphorylated at Thr202 and Thr204 (ERK1) or Thr185 and Thr187 (ERK2) were reduced in Spns2ins/ins rats compared with wild-type and heterozygous rats (Fig. 4, A to C). Both nonphosphorylated and phosphorylated forms of c-Jun were less abundant in the eyelids of Spns2ins/ins rats than in wild-type and Spns2ins/+ rats at E17.5 and E18.5 (Fig. 4, A, D, and E). We also examined the abundances of MEKK1 and JNK, two kinases with roles in eyelid closure (19, 21, 29). MEKK1 was less abundant in the eyelids of Spns2ins/ins rats at E17.5 and E18.5 than in those of the age-matched controls (fig. S7, C and D), but there were no differences in the amount of JNK or phosphorylated JNK in eyelids from the three genotypes at E17.5 and E18.5 (fig. S7, E and F). In primary keratinocytes, the amounts of JNK, phosphorylated JNK, P38, AKT, and phosphorylated AKT did not differ significantly between wild type and Spns2ins/ins, whereas MEKK1 was more abundant in wild-type cells than in Spns2ins/ins cells, consistent with the results in eyelid tissue (fig. S7, G to N). These results indicate that EGFR-ERK and MEKK1–c-Jun signaling is defective in Spns2ins/ins rats.

Fig. 4 S1P rescues defective EGFR-ERK signaling in Spns2ins/ins rats.

(A) Representative Western blot showing EGFR, EGFR phosphorylated at Tyr1068 (pEGFR), ERK1/2, phosphorylated ERK1/2 (pERK1/2), c-Jun, and phosphorylated c-Jun (pc-Jun) in eyelids from WT (+/+), Spns2ins/+ (ins/+), and Spns2ins/ins (ins/ins) rats at E17.5 and E18.5. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) is a loading control. (B to E) Densitometric quantification of pEGFR, pERK1/2, c-Jun, and pc-Jun relative to GAPDH in eyelids from WT, Spns2ins/+, and Spns2ins/ins rats at E17.5 and E18.5. (F) Representative Western blot showing EGFR, pEGFR, ERK1/2, pERK1/2, c-Jun, pc-Jun, and GAPDH in WT and Spns2ins/ins keratinocytes cultured in the absence or presence of S1P. (G to J) Densitometric quantification of pEGFR, ERK1/2, c-Jun, and pc-Jun relative to GAPDH in eyelids from WT and Spns2ins/ins keratinocytes cultured with or without S1P. n = 3 independent keratinocyte cultures for each group. Data in (B) to (E) and (G) to (J) are means ± SEM of three independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001; n.s., not significant.

We next investigated whether S1P could rescue the defects in EGFR signaling in Spns2-deficient keratinocytes. As seen in embryonic eyelid tissue, there was no difference in the amounts of EGFR or ERK1/2 in keratinocytes from wild-type and Spns2ins/ins rats cultured in the presence or absence of S1P (Fig. 4F and fig. S7, O and P). S1P treatment stimulated EGFR phosphorylation in Spns2ins/ins keratinocytes, but not to a level that was comparable to that in untreated wild-type cells (Fig. 4, F and G). S1P treatment stimulated ERK1/2 phosphorylation in Spns2ins/ins keratinocytes to a level comparable to that in untreated wild-type keratinocytes (Fig. 4, F and H). S1P also stimulated ERK1/2 activation in wild-type keratinocytes (Fig. 4, F to H). The amounts of c-Jun and phosphorylated c-Jun were also significantly higher in both wild-type and Spns2ins/ins keratinocytes treated with S1P (Fig. 4, F, I, and J) compared with untreated cells. These results indicate that S1P treatment partially restores EGFR and ERK1/2 signaling and activation of c-Jun in Spns2ins/ins rats.

S1P and EGFR signaling are involved in eyelid closure during development

Eyelids explanted from wild-type rat embryos at E16.5 closed within 24 hours of culture (Fig. 5A). Culturing tissue explants in the presence of the S1PR inhibitors W146 (S1PR1), JTE013 (S1PR2), or VPC23109 (S1PR3) either had no effect on (JTE013) or slowed (W146 and VPC23109) eyelid closure, but closure was complete by 48 hours (Fig. 5, B to D). Cotreatment with all three S1PR inhibitors prevented eyelid closure (Fig. 5E). When tissue explants were treated with the S1PR inhibitor cocktail for 2 hours and then with S1P for another 46 hours, the eyelids did not close completely (Fig. 5F). However, when explants were treated with all three S1PR inhibitors for 2 hours and then with EGF for another 46 hours, the eyelids closed and sealed (Fig. 5G).

Fig. 5 S1PR or EGFR inhibition prevents closure of WT embryonic eyelid explants in vitro.

(A to E) Eyelid explants from WT embryos treated with vehicle (Ctrl), the S1PR1 inhibitor W146, the S1PR2 inhibitor JTE013, the S1PR3 inhibitor VPC23109, or all three inhibitors simultaneously (WJV). Arrows indicate sites of eyelid fusion. Dotted lines highlight open eyelids (n = 5 rats for each group). (F and G) WT eyelid explants were treated with WJV for 2 hours, followed by S1P or EGF for an additional 46 hours (n = 5 rats for each group). (H to K) WT eyelid explants were treated with the EGFR inhibitors AG1478, gefitinib, erlotinib, or PD-153053 (n = 5 rats for each group). (L and M) WT eyelid explants were treated with gefitinib for 2 hours, followed by EGF or S1P for an additional 46 hours (n = 5 rats for each group). Scale bars, 200 μm.

Inhibiting EGFR signaling also inhibited eyelid closure in tissue explants from wild-type embryos. Eyelids failed to close within 48 hours of culture in the presence of the EGFR inhibitors AG1478, gefitinib, erlotinib, or PD-153053 (Fig. 5, H to K). After gefitinib treatment for 2 hours and then with EGF or S1P for another 46 hours did not restore eyelid closure; although EGF partially restored closure, the leading edges did not fuse, causing the lids to reopen (Fig. 5, L and M). Taken collectively, these results demonstrate that both S1PR and EGFR signaling are important for eyelid closure in developing rats and suggest that S1PR signaling may be upstream of EGFR signaling in this context.

S1PR signaling occurs upstream of EGFR signaling in keratinocytes

Phosphorylation of EGFR and that of ERK1/2 are commonly used as indicators of EGFR activation (67). S1PR inhibitors reduced the amount of phosphorylated EGFR in untreated, vehicle-treated, and S1P-treated wild-type keratinocytes and had no effect on the ability of EGF to induce EGFR phosphorylation (Fig. 6, A and B), consistent with S1PR signaling being upstream of EGFR signaling. S1PR inhibitors did not affect the amount of phosphorylated ERK1/2 in vehicle-treated wild-type keratinocytes and did not prevent EGF from inducing ERK1/2 phosphorylation (Fig. 6, A and C). However, compared with control cells, S1P treatment had little or no effect on the amount of phosphorylated EGFR but increased the amount of phosphorylated ERK1/2 (Fig. 6, A and C). Phosphorylation of EGFR depended on the presence of ligand, but phosphorylation of ERK1/2 did not, suggesting that S1P induces phosphorylation of ERK1/2 through both EGFR-dependent and EGFR-independent mechanisms. Although S1P did not induce EGFR phosphorylation, S1PR inhibitors reduced basal phosphorylation of EGFR in wild-type keratinocytes, but this could be reversed by EGF.

Fig. 6 S1PR acts upstream of EGFR signaling in WT keratinocytes.

(A) Representative Western blot showing EGFR, EGFR phosphorylated at Tyr1068 (pEGFR), ERK1/2, and phosphorylated ERK1/2 (pERK1/2) in WT keratinocytes treated with the indicated combinations of dimethyl sulfoxide (DMSO); methanol; the S1PR inhibitors W146, JTE013, and VPC23109 (WJV); S1P; and EGF. GAPDH is a loading control. (B and C) Densitometric quantification of pEGFR and pERK1/2 relative to GAPDH in WT keratinocytes treated as indicated. (D) Representative Western blot showing EGFR, pEGFR, ERK1/2, pERK1/2, and GAPDH in WT keratinocytes treated with DMSO, methanol, gefitinib, S1P, and EGF as indicated. (E and F) Densitometric quantification of pEGFR and pERK1/2 relative to GAPDH in WT keratinocytes treated as indicated. n = 3 independent keratinocyte cultures for each group. Data in (B), (C), (E), and (F) are means ± SEM of three independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001; n.s., not significant.

The amounts of pshosphorylated EGFR and phosphorylated ERK1/2 in wild-type keratinocytes increased in cells treated with EGF and decreased in cells treated with the EGFR inhibitor gefitinib (Fig. 6, D to F). Neither S1P nor EGF increased the abundance of phosphorylated EGFR and phosphorylated ERK1/2 after gefitinib treatment (Fig. 6, D to F). These results are consistent with S1PR stimulating EGFR signaling during eyelid development in rats.

ADAM10 and ADAM17 mediate S1PR-induced activation of EGFR

Because GPCR activation can lead to EGFR activation by inducing the cleavage of HB-EGF by ADAM proteases (68), we investigated whether activation of S1PRs stimulated ADAM-dependent cleavage of HB-EGF. Treatment of wild-type keratinocytes with the nonspecific metalloprotease inhibitor BB94 reduced the amounts of phosphorylated EGFR and phosphorylated ERK1/2 in wild-type keratinocytes (Fig. 7, A to C). S1P did not prevent the BB94-induced reduction in EGFR and ERK1/2 phosphorylation, but EGF did (Fig. 7, A to C), indicating that metalloproteases are involved in S1PR-induced activation of EGFR in keratinocytes.

Fig. 7 ADAM10 and ADAM17 are involved in S1PR-mediated transactivation of EGFR.

(A) Representative Western blot showing EGFR, EGFR phosphorylated on Tyr1068 (pEGFR), ERK1/2, and phosphorylated ERK1/2 (pERK1/2) in WT keratinocytes treated with the indicated combinations of DMSO, methanol, the nonspecific metalloprotease inhibitor BB94, S1P, and EGF. GAPDH is a loading control. (B and C) Densitometric quantification of pEGFR and pERK1/2 relative to GAPDH in WT keratinocytes treated as indicated. (D) Representative Western blot showing EGFR, pEGFR, ERK1/2, pERK1/2, and GAPDH in WT keratinocytes treated with DMSO, methanol, the ADAM10 inhibitor GI254023X, S1P, and EGF as indicated. (E and F) Densitometric quantification of pEGFR and pERK1/2 relative to GAPDH in WT keratinocytes treated as indicated. (G) Representative Western blot showing EGFR, pEGFR, ERK1/2, pERK1/2, and GAPDH in WT keratinocytes treated with DMSO, methanol, the ADAM17 inhibitor TAPI-2, S1P, and EGF as indicated. (H and I) Densitometric quantification of pEGFR and pERK1/2 relative to GAPDH in WT keratinocytes treated as indicated. n = 3 independent keratinocyte cultures for each group. Data in (B), (C), (E), (F), (H), and (I) are means ± SEM of three independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001; n.s., not significant.

Although several ADAM proteases can cleave EGFR ligands, ADAM10 and ADAM17 are the major mammalian EGFR ligand sheddases (67, 69). ADAM10 is the main sheddase of EGF and betacellulin, whereas ADAM17 is the major convertase of HB-EGF, TGF-α, amphiregulin, and epiregulin (67). We treated wild-type keratinocytes with GI254023X and TAPI-2, which are specific inhibitors of ADAM10 and ADAM17, respectively. Phosphorylated EGFR and phosphorylated ERK1/2 were decreased in wild-type keratinocytes treated with GI254023X (Fig. 7, D to F) or TAPI-2 (Fig. 7, G to I) compared with control cells. Treatment with EGF, but not with S1P, reversed the inhibitory effect of both GI254023X and TAPI-2 on EGFR and ERK1/2 phosphorylation (Fig. 7, D to I). In contrast, MMP-2 (matrix metalloprotease 2)/MMP-9 inhibitor I, an inhibitor of MMP-2 and MMP-9, did not prevent EGFR and ERK1/2 phosphorylation in response to S1P or EGF (fig. S8, A to C). These results indicate that S1P stimulates EGFR signaling in keratinocytes through a mechanism that depends on ADAM10 and ADAM17.

ADAM10 and ADAM17 activity is statistically significantly lower in Spns2ins/ins keratinocytes than in wild-type cells

The pro and mature forms of ADAM10 and ADAM17 were present in similar amounts in wild-type and Spns2ins/ins keratinocytes cultured in the absence or presence of S1P (Fig. 8, A and B, and fig. S9, A to F). We measured ADAM10 and ADAM17 activity in lysates of eyelid tissue explants and keratinocytes from wild-type and Spns2ins/ins rat using a commercially available kit. There was no difference in ADAM10 activity in eyelid tissue explants from wild-type and Spns2ins/ins embryos at E15.5 to E18.5 yet (Fig. 8C). However, ADAM17 activity was lower in explants from Spns2ins/ins embryos at E16.5 and E17.5, a critical time point of eyelid closure (Fig. 8D). In cultured keratinocytes, both ADAM10 and ADAM17 activities were lower in Spns2ins/ins rats (Fig. 8, E and F). The enzymatic activity of ADAM17 was ~10-fold higher than that of ADAM10 in eyelid tissue explants and keratinocytes (Fig. 8, C to F). Treatment of wild-type embryonic eyelid tissue explants with the ADAM10 inhibitor GI254023 or the ADAM17 inhibitor TAPI-2 completely prevented eyelid closure (Fig. 8G). The results indicate that ADAM10 and ADAM17 play an important role in eyelid closure and that Spns2 deficiency reduces ADAM10 and ADAM17 activity.

Fig. 8 ADAM10 and ADAM17 are involved in the EOB phenotype.

(A and B) Representative Western blots showing ADAM10 and ADAM17 in WT and Spns2ins/ins keratinocytes treated with vehicle or S1P. P, pro form; M, mature form. GAPDH is a loading control (n = 3 independent keratinocyte cultures per group). (C and D) ADAM10 and ADAM17 activity in eyelid tissue explants from WT (+/+) and Spns2ins/ins (ins/ins) embryos at E15.5 to E18.5. (E and F) ADAM10 and ADAM17 activity in keratinocytes from WT and Spns2ins/ins embryos. Data in (C) to (F) are means ± SEM of three independent experiments. *P < 0.05; **P < 0.01. (G) Eyelid closure assay in eyelid explants from WT embryos treated with vehicle (Ctrl), the ADAM17 inhibitor TAPI-2, or the ADAM10 inhibitor GI254023X. Arrows indicate sites of eyelid fusion (n = 5 rats for each group). Scale bar, 400 μm.

Inactive Rhomboid protein 1 (iRhom 1) and iRhom2 are key regulators of ADAM17 activity (70). There were no difference in the amounts of iRhom1 and iRhom2 present in wild-type and Spns2ins/ins keratinocytes cultured in the presence or absence of S1P (fig. S9, G to I), indicating that the reduced ADAM17 activity in Spns2ins/ins keratinocytes is not due to a reduction in iRhom1 or iRhom2.

Timp3 expression is increased in Spns2ins/ins eyelid tissue explants and cultured keratinocytes

Tissue inhibitors of metalloproteinases (Timps) are physiological inhibitors of MMPs and ADAMs in the extracellular environment (71). Although Timp3 is an inhibitor of ADAM17, it can also effectively inhibit ADAM10 activity (72). The abundance of Timp3 mRNA was ~2-fold higher in Spns2ins/ins keratinocytes than in wild-type cells (Fig. 9A). By contrast, there was no difference in Timp4 expression between wild-type and Spns2ins/ins keratinocytes (Fig. 9A). The Timp3 protein was also more abundant in Spns2ins/ins keratinocytes than in wild-type cells (Fig. 9, B and C). Spns2ins/ins keratinocytes stained strongly for Timp3 (Fig. 9D), and immunoblotting showed that Timp3 was more abundant in eyelid tissues from Spns2ins/ins embryos at E16.5, E17.5, and E18.5 than in tissues from wild-type embryos (Fig. 9, E and F). Immunohistochemical staining also showed that Timp3 was virtually undetectable in the eyelids of wild-type embryos but highly abundant in epithelial cells of both the conjunctiva and epidermis in the eyelids of Spns2ins/ins embryos (Fig. 9G). Furthermore, lipotransfection of Timp3 prevented eyelid closure in wild-type eyelid explants (Fig. 9H), indicating that Spns2 deficiency promotes Timp3 expression and activity in developing eyelids.

Fig. 9 Timp3 mRNA and protein are increased in Spns2ins/ins rats.

(A) Timp3 and Timp4 expression relative to one of the WT groups in WT (+/+) and Spns2ins/ins (ins/ins) keratinocytes as measured by qPCR (n = 4 independent keratinocyte cultures per group). (B) Representative Western blot showing TIMP3 in keratinocytes from WT and Spns2ins/ins rats (n = 3 independent keratinocyte cultures per group). GAPDH is a loading control. (C) Densitometric quantification of Timp3 relative to GAPDH level in WT and Spns2ins/ins rats. (D) Immunostaining showing Timp3 in keratinocytes from WT and Spns2ins/ins rats (n = 3 rats of each genotype, six wells of cells from each harvest). Scale bar, 40 μm. (E) Representative Western blot showing TIMP3 in embryonic eyelid tissue explants from WT and Spns2ins/ins rats at E16.5, E17.5, and E18.5 (n = 3 rats per group). (F) Densitometric quantification of TIMP3 relative to GAPDH in embryonic eyelid tissue explants from WT and Spns2ins/ins rats at E16.5, E17.5, and E18.5 (n = 3 rats per group). (G) Immunofluorescence showing TIMP3 in embryonic eyelid tissue explants from WT and Spns2ins/ins rats (n = 4 rats per group). Scale bar, 50 μm. (H) Eyelid tissue explants from WT embryos were lipofected with empty vector or Timp3 at E16.5 in vitro (n = 5 rats per group). Arrows indicate sites of eyelid fusion. Scale bar, 400 μm. Images are representative. Data in (A), (C), and (F) are means ± SEM of three independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001.

MAL localizes to the cytoplasm in Spns2ins/ins keratinocytes, and miR-21 and miR-222 mimics function in eyelid closure

SRF, one of the best studied MADS box transcriptional factors that recognizes the CArG box [CC(AT)6GG] DNA sequences in target gene promoters, inhibits TIMP3 expression in human lung fibroblasts (73, 74). In addition, keratinocyte-specific ablation of Srf in mice caused the EOB phenotype with a marked loss of keratinocyte stress fibers, similar to what we have observed in Spns2ins/ins rats (75, 76). By Western blot, there were no differences between wild-type and Spns2ins/ins keratinocytes in the amount of SRF, the amount or phosphorylation of the SRF cofactor Elk1, or the amount of cFOS, which is transcriptionally regulated by Elk1 and SRF (Fig. 10A and fig. S10, A, C, and D). However, the abundance of MAL, another SRF cofactor that binds to G-actin in the cytoplasm and prevents F-actin polymerization, was lower in Spns2ins/ins keratinocytes than in wild-type cells (Fig. 10, A and B). S1P treatment had no effect on the abundance of SRF or phosphorylated Elk1 but restored the abundance of MAL in Spns2ins/ins cells so that it was similar to that in wild-type cells (Fig. 10, A and B). The abundance of Mig6, a negative regulator of the EGFR-MAPK (mitogen-activated protein kinase) cascade that is transcriptionally activated by MAL and SRF (77), was lower in Spns2ins/ins keratinocytes than in wild-type cells (fig. S10, E and F). Taken collectively, these results indicate that MAL-SRF function is abnormal in Spns2ins/ins keratinocytes.

Fig. 10 MAL localizes to the cytoplasm in Spns2ins/ins keratinocytes.

(A) Representative Western blot showing MAL in primary WT (+/+) and Spns2ins/ins (ins/ins) keratinocytes treated with vehicle or S1P. GAPDH is a loading control. Arrowhead indicates MAL (n = 3 rats per group). (B) Densitometric quantification of MAL relative to GAPDH in untreated and S1P-treated keratinocytes from WT and Spns2ins/ins rats. (C) Distribution of MAL in WT keratinocytes. Cyt, cytoplasmic; C + N, both cytoplasmic and nuclear; Nuc, nuclear (n = 6 wells, each from three separate harvests). Scale bar, 20 μm. (D) Subcellular distribution of MAL in primary keratinocytes from WT and Spns2ins/ins rats treated as indicated or not treated (CTL) (n = 6 wells, each from three separate cultures). (E) Fraction of WT and Spns2ins/ins keratinocytes in which MAL was present in the nucleus after the indicated treatments (n = 6 rats per group). Statistical analyses: *, versus CTL WT group. #, versus CTL Spns2ins/ins group. (F to H) miR-21 and miR-221, and miR-222 expression in keratinocytes from WT and Spns2ins/ins rats by qPCR (n = 5 rats per group). (I) Rescue experiments in which the miR-21, miR-221, or miR-222 mimics were lipotransfected into Spns2ins/ins embryos singly and in the indicated combinations (n = 5 rats per group). Scale bar, 500 μm. Images are representative. Data in (B) and (E) to (H) are means ± SEM of three independent experiments. *P < 0.05; **P < 0.01; ##P < 0.01; ###P < 0.001.

Actin polymerization promotes the translocation of MAL into the nucleus, and this process involves RhoA, as well as the ROCK–LIM kinase (LIMK)–cofilin and mDia (mammalian Diaphanous-related formin 1) pathways (7880). We examined the abundance of several molecules involved in actin-mediated control of MAL activity, including RhoA, ROCK1, ROCK2, LIMK1, LIMK2, phosphorylated LIMK1 and LIMK2, cofilin, phosphorylated cofilin, and mDia, but found that there was no significant difference in the amounts of these molecules between wild-type and Spns2ins/ins keratinocytes (fig. S10G). MAL exhibited three patterns of localization—cytoplasmic, nuclear, or both cytoplasmic and nuclear—depending on its activation (Fig. 10C). Whereas MAL exhibited nuclear localization in the majority of wild-type keratinocytes, it was present in both the cytoplasm and nucleus in most Spns2ins/ins cells, leading us to conclude that either S1P or EGF signaling promoted the nuclear accumulation of MAL (Fig. 10D). Consistent with this, treatment with S1PR inhibitors or gefitinib reduced MAL nuclear localization in wild-type keratinocytes, and treatment with S1P or EGF increased MAL nuclear localization in Spns2ins/ins keratinocytes (Fig. 10, D and E). In addition, treating wild-type keratinocytes with gefitinib followed by S1P failed to reverse the nuclear location of MAL (Fig. 10, D and E). Our results show that MAL localized predominantly to the nucleus in wild-type keratinocytes but to both the cytoplasm and nucleus in Spns2ins/ins cells and that S1P promoted the nuclear accumulation of MAL in a manner that depended on the activation of EGFR signaling.

miR-21 and miR-222 are repressed in Spns2ins/ins keratinocytes

Timp3 is also regulated posttranscriptionally in various cell types by several miRNAs, including miR-21, miR-221, and miR-222 (80, 81). The miR-21 promoter contains one CArG box, and the promoter of the cluster containing miR-221 and miR-222 has an activator protein–1 (AP-1)–binding site. MAL-SRF promotes the expression of miR-21, and c-Jun promotes the expression of miR-221 and miR-222 (81, 82). qPCR showed that expression of miR-21, miR-221, and miR-222 was significantly lower in Spns2ins/ins keratinocytes than in wild-type cells (Fig. 10, F to H). Lipofection of miR-21 or miR-222 mimics into Spns2ins/ins embryos in utero nearly completely rescued eyelid closure (Fig. 10I), whereas the miR-221 mimic only weakly promoted eyelid closure (Fig. 10I). A mixture of miR-221 and miR-222 mimics induced partial eyelid closure, except that a small area of the eye remained open, similar to that observed after injection of the miR-222 mimic alone (Fig. 10I). Coinjection of all three miRNA mimics completely rescued eyelid closure in Spns2ins/ins embryos (Fig. 10I). These results indicate that both miR-21 and miR-222 are indispensable for eyelid closure and that misregulation of these miRNAs contribute to the EOB phenotype in Spns2ins/ins rats.

DISCUSSION

FGF10 is required for eyelid closure and is produced by eyelid mesenchymal cells as early as E11.5 in mice, but its receptor, FGFR2, is present on epithelial cells (83, 84). It promotes the early stages of eyelid formation by stimulating epithelial cell proliferation and maintains the migration of epithelial cells at the leading edge (83). During the late stages of eyelid closure, FGF10 promotes the production of TGF-α and the TGF-β family member activin B (83). FGF10 failed to rescue the rat Spns2ins/ins EOB phenotype in eyelid tissue explants in vitro or in amniotic fluid injections in vivo (Fig. 3G). We suggest that the role of FGF10 in eyelid closure may depend on S1P signaling or on the activity of ADAM10 or ADAM17, or both. Maretzky et al. (85) showed that FGF10 binds to FGFR2b to stimulate EGFR signaling by promoting ADAM17 activity during keratinocyte migration. Nevertheless, the relationship between FGF10, S1PR activity, ADAM activity, and cell migration requires further investigation.

Herr et al. (11) reported no change in EGFR phosphorylation, but a 3.8-fold increase in ERK1/2 activation, in S1P-treated wild-type mouse embryonic fibroblasts (MEFs). These changes were also observed in Spns2-deficient keratinocytes treated with S1P here (Figs. 6, A and D, and 7, A, D, and G). This discrepancy may be due to the fact that under normal conditions, wild-type keratinocytes and fibroblasts release an amount of endogenous S1P that is sufficient to activate S1PR1 to S1PR3 and stimulate EGFR signaling (50). Here, we found that 1 μM S1P increased ERK1/2 activation, but not EGFR phosphorylation, in wild-type keratinocytes, suggesting that S1PR1 to S1PR3 activation can activate several MAPK molecules, including ERK1/2, independent of EGFR signaling. Furthermore, S1PR1 to S1PR3 inhibitors impeded the activation of ERK1/2 in wild-type keratinocytes, consistent with the results from S1pr2/3-null MEFs (Fig. 6, A and C, lanes 4 and 8, respectively) (11).

The paradigm of triple membrane passing signaling (TMPS) is a widely accepted model of receptor tyrosine kinase (RTK) transactivation by GPCR stimulation (86). It has been shown that EGFR transactivation by GPCRs is a general phenomenon that occurs in diverse cell types through various GPCRs (86, 87). In addition to HB-EGF, other EGFR ligands, such as androgen receptor and TGF-α, are also involved in EGFR transactivation. S1P can also transactivate several RTKs, including EGFR and platelet-derived growth factor receptor signaling, through GPCRs such as S1PR3 (8890). In the TMPS mechanism of GPCR-induced EGFR activation, ERK-MAPK activation is induced by various EGFR ligands that are shed from the plasma membrane (87). SIPR2 and S1PR3 were critical for eyelid closure and attenuated EGF activity (11). Although many genes are involved in EOB, only a few cause complete penetrance of the EOB phenotype when knocked out, such as egfr and Adam1, Spns2, and S1pr2/3 (11). These genes were included in the S1P-EGFR axis we studied, confirming that EGFR transactivation by S1P is fundamental for eyelid development.

Six EGFR ligands are cleaved by multiple ADAMs, including ADAMs 9, 10, 12, 15, 17, and 19. Among these, two membrane-anchored metalloproteinases, ADAM10 and ADAM17, play central roles in the regulation of EGFR signaling (67, 91). Although ADAM17 and ADAM10 share several common EGFR ligands, the results of in vivo and in vitro experiments are often divergent. ADAM10 as an alternative sheddase for ADAM17 substrates is only relevant when Adam17 is deleted or inhibited in in vitro assays (92). However, ADAM10 and ADAM17 cannot compensate for each other’s loss in vivo (93, 94). All mouse models of Adam17 loss of function—hypomorphic mutants, knockouts, and epidermal-specific deletion—develop EOB and EGFR signaling disruption, although Adam10 function is intact (9395). In addition, ADAM10 is critical for early embryogenesis because it cleaves Notch, which is involved in the development of several organs (91). Adam10−/− mice die at E9.5 with malformed yolk sac vessels due to defective Notch signaling (96). Epidermal-specific Adam10 deletion leads to perinatal lethality, barrier impairment, and the absence of sebaceous glands, without an apparent EOB phenotype (97). Although in vitro data support the roles of both ADAM10 and ADAM17 in the regulation of Notch shedding, Adam10 deletion in the epidermis did not affect ADAM17 expression and activity and failed to rescue Notch1 shedding (97). Presently, the EGFR ligands that induce EOB have only been reported in TGF-α–null and HB-EGF–null mutant mice (3, 4, 64). TGF-α and HB-EGF are ADAM17 substrates (67). Although ADAM10 was involved in eyelid fusion in vitro, our results show that ADAM17 was the main enzyme that cleaved EGFR ligands during eyelid closure in developing rats (98). We also found that inhibition of either ADAM10 or ADAM17 impeded the activation of EGFR and ERK1/2 in cultured wild-type keratinocytes. Further studies are needed to determine whether the inhibition of ADAM10 or ADAM17 can affect the activity of other sheddases in S1P-treated keratinocytes.

Timp3, which inhibits ADAM17 by chelating extracellular Zn2+ at its N terminus (99, 100), was more abundant in Spns2-deficient keratinocytes than in wild-type keratinocytes. Timp3 has a broader inhibitory profile than other Timps and can inhibit ADAM10, ADAM12, and several a disintegrin and metalloproteinase with thrombospondin motifs (ADAMTSes) in addition to inhibiting ADAM17 (69, 101). Timp3-deficient mice display increased soluble TNF accompanied by inflammation in several tissues (102, 103). Although the detailed mechanism of the Timp3 inhibition of ADAM10 and ADAM17 is not well elucidated, several studies have shown that Timp3 is secreted into the extracellular environment and binds to and inhibits ADAM10 and ADAM17 (104). Timp3 is regulated by the transcriptional factor AP-1, a heterodimer formed by members of the c-Jun and cFos families (105). Garofalo et al. (81) showed that knockdown of c-Jun, but not cFos, reduced miR-222 expression by 40 to 45%. Furthermore, studies with c-Jun and cFos single knockouts and cJun and JunB double knockouts have shown that c-Jun–mediated expression of miR-22 reduces Timp activity, thereby promoting ADAM17 activity (63, 105, 106). In addition, c-Jun promotes the formation of the epidermal leading edge by stimulating an EGFR autocrine loop, which supports its role in the EOB phenotype (17, 18). We therefore speculate that c-Jun induces miR-222 to decrease Timp3 during eyelid closure, thus inhibiting ADAM10 and ADAM17 activity and EGFR signaling.

In Spns2-deficient eyelid tissues, expression of miR-21, an oncomiR with critical targets in malignant melanoma, was also decreased (107, 108). Timp3 is also an important target of miR-21 in skin development (107, 109). Recently, Li et al. (82) reported that mesenchymal stem cells exhibit high miR-21 expression when cultured on stiff substrates. The miR-21 promoter contains one conserved CArG box, which potentially binds to SRF-MAL. Our experiments are consistent with SRF-MAL, promoting the expression of miR-21. In chromatin immunoprecipitation assays, antibodies recognizing SRF precipitated a > 5-fold higher amount of miR-21 than did assays with the control antibody. Overexpression of full-length MAL resulted in a 6-fold increase in miR-21 expression, whereas knockdown of MAL resulted in a 1.5-fold decrease (82). In this study, we found that miR-222 and miR-21 had synergistic effects on eyelid closure and completely rescued the EOB phenotype in Spns2-deficient embryos. This may be due to the fact that Timp3 is targeted by both miR-21 and miR-222.

Reduced F-actin polymerization and reduced migration of epidermal cells are common characteristics of the EOB phenotype (18, 64). EGFR promotes the activities of Rho and its two targets ROCK and mDia1 in primary keratinocytes from embryonic mice (22, 23). ROCK promotes F-actin stabilization by phosphorylating LIMK, which in turn phosphorylates and inactivates the actin depolymerizing factor cofilin, whereas mDia1 can induce de novo F-actin assembly (79, 80). The guanosine triphosphatase RhoA activates ROCK1/2 to stimulate actin stress fiber formation and actin polymer stabilization, which promotes epithelial cell migration in eyelids. ROCK1/2 and LIMK2 knockout mice display the EOB phenotype with reduced actin stress fiber formation in keratinocytes (22, 23, 110). MAL binds to SRF to mediate RhoA-actin signaling in cells (78, 111). SRF responds to two distinct signaling pathways, the ternary complex factors (TCFs) or the myocardin-related transcription factors (MTRFs) (78, 111). Activation of the first pathway has been demonstrated for EGFR ligands and that of the second pathway has been demonstrated for GPCR ligands, including LPA and S1P, but not for EGFR ligands (77, 112). In this study, we found that S1P effectively induced stress fiber formation and enhanced MAL accumulation in the nucleus. Although we reported that this process was largely dependent on EGFR signaling, the precise mechanism of EGFR-Rho signaling in eyelid epithelial migration during embryogenesis has not been elucidated. Because both S1P and EGFR signaling and SRF-MAL are disrupted in EOB, either TCF or MTRF mediated SRF signaling, or both.

Our results can be summarized in the following model (Fig. 11): In normal keratinocytes, Spns2 controls the release of S1P from the cytoplasm to the outside of the cell, where it activates its receptors, in particular S1PR1 to S1PR3, which can transactivate EGFR signaling by stimulating ADAM10- and ADAM17-mediated shedding of EGFR ligands. Activated EGFR may contribute to eyelid closure by activating c-Jun and promoting the nuclear accumulation of MAL to enhance F-actin polymerization and stress fiber formation in keratinocytes. c-Jun and MAL may posttranscriptionally inhibit TIMP3 and thus enhance ADAM10 and ADAM17 activity, by stimulating the expression of miR-21 or miR-222, or both. In Spns2-deficient rats, the reduced release of S1P disrupts this network, causing the EOB phenotype.

Fig. 11 Schematic diagram of the network through which S1P network transactivates EGFR and stimulates eyelid closure in developing rats.

Dashed arrows indicate pathways reported by others; red arrows indicate the stimulation or inhibition of relevant molecules in Spns2ins/ins rats; red question marks indicate pathways that remain unstudied.

MATERIALS AND METHODS

Animal studies

The transgenic GR expressing exogenous Egfp (SD-Tg CAG-EGFP) was generated by O. Masaru (Osaka University), acquired from Japan SLC Inc., and maintained in our SPF animal facility. All experimental procedures were approved by the Institutional Animal Care and Use Committee of South China University of Technology and Fourth Military Medical University.

Inverse PCR, qPCR, and miRNA quantification

For inverse PCR, genomic DNA was extracted from ~5 g of fresh liver from an Spns2ins/ins rat and immediately frozen in liquid nitrogen. The tissue was crushed, added to 5 ml of lysis buffer containing tris-Cl (20 mM), EDTA (5 mM), NaCl (400 mM), SDS (1%), and proteinase K (100 μg/ml) and incubated for 3 hours at 50°C. The DNA was isolated by two rounds of phenol extraction, precipitated with ethanol, pelleted, and washed twice with 70% ethanol. The DNA was dissolved in 1 ml of tris-EDTA buffer (pH 8.0) and stored at 4°C. About 5 μg of Spns2ins/ins genomic DNA was digested with Bgl II or Eco RI (Takara Biomedicals) overnight at 37°C, and the DNA fragments were purified with the SanPrep Column PCR Product Purification Kit (Sangon Biotech). For recircularization, 0.1 μg of the digested DNA fragment was diluted to a concentration of 0.005 μg in ligation buffer. T4 DNA ligase (Takara Biomedicals) was added to a concentration of 1 U/μl, and the DNA was incubated overnight at 16°C. Nested PCR was carried out with the recircularized DNA sample as the template. Primers P1 and P2 were used for the first round of PCR. The products were purified and used as templates for the second round of PCR with primers P1 and P3. The product from the second round of PCR was used as the template for the third round of PCR with primers P3 and P4. The product from the third round of PCR was purified for sequencing.

For qPCR, total RNA was extracted from keratinocytes from wild-type and Spns2ins/ins rats using RNAiso Plus (Takara Biomedicals). Complementary DNA (cDNA) was synthesized using the PrimeScript RT Master Mix (Takara Biomedicals), and qPCR was performed using the SYBR Premix Ex Taq II PCR Kit (Takara Biomedicals) and the Bio-Rad CFX96 Thermal Cycler.

miRNA quantification was performed using Stem-Loop RT-PCR as previously described (113). The purification of small RNAs was performed with the mirPremier microRNA Isolation Kit (Sigma-Aldrich), according to the manufacturer’s protocol. Stem-Loop RT-PCR consisted of two steps: RT and qPCR. First, the Mir-21RT, Mir-221RT, Mir-222RT, and U6-2 primers were used to perform RT with the RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific). The expression of miRNAs was then quantified by qPCR using SYBR Premix Ex Taq II (Takara Biomedicals) and Mir-21s + ASP, Mir-221s + ASP, Mir-222s + ASP, or U6-1 + U6-2 primers (table S1).

Southern blotting

Genomic DNA (~2.5 to 5 μg) was digested with Bgl II and Pst I, electrophoresed on an 0.8% agarose gel at 25 V for ~18 hours, blotted onto a BrightStar-Plus Positively Charged Nylon Membrane (Thermo Fisher Scientific) in 0.4 M NaOH and 0.6 M NaCl transfer solution, and then baked at 80°C for 2 hours. The membranes were hybridized overnight at 42°C using an Eg fp fragment labeled by DIG DNA Labeling and Detection Kit (Roche) as the probe. The membranes were washed at 42°C in 0.1× saline sodium phosphate EDTA and 0.1% sodium dodecyl sulfate (Roche) until the background signal was eliminated. The signal was detected with the DIG DNA Labeling and Detection Kit (Roche), and images were captured using a Bio-Rad ChemiDoc XRS+ System.

Whole-genome sequencing

WGS was carried out by the Shanghai Biotechnology Corporation. Genomic DNA was extracted from the epidermis of GR rats using the DNeasy Blood and Tissue Kit (Qiagen). The genomic library was prepared using the NEBNext Ultra DNA Library Prep Kit according to the manufacturer’s instructions. In brief, sonicated DNA was end repaired into dA-tailed fragments. The NEBNext adaptor (New England Biolabs) was ligated to the DNA fragments. The adaptor-ligated DNA was purified with AMPure XP beads (Beckman Coulter) and subsequently amplified using index and universal PCR primers provided in the NEB Multiplex Oligo Kit (New England Biolabs). The PCR products were then purified with AMPure XP beads. The genomic library concentration was measured with a Qubit 2.0 Fluorometric Quantitation System (Thermo Fisher Scientific), and the size distribution was generated by the Agilent 2100 Bioanalyzer System (Agilent). For Illumina sequencing, the high-quality genomic libraries were used for 2× 150-bp paired-end sequencing on the Illumina HiSeq X-Ten System (Illumina). FermiKit was used to assemble the reads into unitags, map them to the rat genome, and find the insertional site of exogenous Egfp in the rat genome. Genome sequencing has been deposited into the Sequence Read Archive (SRA) of the National Center for Biotechnology Information (NCBI) (www.ncbi.nlm.nih.gov/sra) with the accession number SRR7415781.

In vivo electroporation

Wild-type and Spns2ins/ins 16.5-day pregnant rats were anesthetized with 1% pentobarbital and positioned on a heating pad for laparotomy. Spns2 plasmid or the negative control was injected into the amniotic sac. Embryo was held with forceps-type electrodes, paralleled to the embryonic anteroposterior axis. Three to five electric pulses were delivered by an electroporator every second with a duration of 50 ms. The status of eyelid closure was examined under an Olympus STZ16 stereomicroscope after 4 days, and images were acquired with an Olympus MVX10 macroview microscope. At least three embryos were used for each treatment.

Giemsa staining

Blood of wild-type, Spns2ins/+, and Spns2ins/ins adult rats were obtained from the tail vein. The blood slides were fixed in methanol for 2 to 3 min before immersing them in freshly prepared 5% Giemsa’s solution for 20 to 30 min. The slides were flushed with tap water and air dried. The counts of lymphocyte, monocyte, neutrophil, eosinophil, and basophil were obtained under an Olympus BX53 fluorescent microscope. The percentage of each cell type was calculated.

Rat eyelid primordial tissue explant cultures

Eyelid tissue explants were cultured as previously described (83). In brief, eyelids from embryos at E16.5 were isolated with microscissors under an Olympus STZ16 stereomicroscope and cultured in 24-well plates containing 150 μl of keratinocyte media 2 (KM2) supplemented with EGF (100 ng/ml), TGF-β1 (50 ng/ml, PeproTech), HB-EGF (50 ng/ml, PeproTech), S1P (10 μM), or liposome 2000 combined with the vehicle or Timp3 plasmid at 37°C in a humidified atmosphere of 5% CO2 for 48 hours. For EGFR and S1PR1 to S1PR3 inhibition experiments, eyelid tissue explants were cultured in the presence of AG1478 (2 μM, Tocris Bioscience), PD153035 (5 μM, Tocris Bioscience), gefitinib (10 μM), W146 (10 μM), VPC23109 (10 μM), or JTE013 (10 μM) for 48 hours.

Isolation and culture of primary keratinocytes

Primary keratinocytes were isolated and cultured as previously described (17). In brief, 3-day postnatal mice were sacrificed by CO2 asphyxiation. The bodies were rinsed in 75% ethanol and dried with sterile gauze. The mouse was cut from the nose to the tail, and the skin was removed in one piece. The skin was rinsed several times in washing solution [PBS (pH 7.4), containing penicillin (100 U/ml), streptomycin (100 μg/ml), fungizone (0.25 μg/ml), and gentamicin (50 μg/ml)] and incubated in medium containing dispase [dispase II (5 U/ml; Sigma-Aldrich), streptomycin (100 μg/ml), fungizone (0.25 μg/ml), and gentamicin (50 μg/ml)] overnight at 4°C. The dermis was separated from the epidermis with forceps, minced, and digested in 0.25% trypsin for 10 min. Rat keratinocyte culture medium containing defined KM2 (PromoCell) without EGF was used. Keratinocytes were plated into 24- or 6-well plates at previously established cell densities. Media were changed every 2 days. The cells were trypsinized, and three independent batches of cells were counted.

For stimulation experiments, 1 μM S1P (Cayman Chemical) or EGF (100 ng/ml; PeproTech) was added into KM2 and cells were treated for 30 min. In some cases, cells were pretreated with inhibitors (10 μM) for 30 min, followed by S1P (1 μM) or EGF (100 ng/ml), and incubated for another 30 min. The inhibitors were as follows: gefitinib (EGFR inhibitor, Tocris Bioscience), AG1478 (EGFR inhibitor, Sigma-Aldrich), erlotinib (EGFR inhibitor, Sigma-Aldrich), and PD-153053 (EGFR inhibitor, Tocris Bioscience); W146 (S1PR1 inhibitor, Tocris Bioscience), JTE013 (S1PR2 inhibitor, Tocris Bioscience), and VPC23109 (S1PR3 inhibitor, Tocris Bioscience); and BB94 (nonspecific metalloprotease inhibitor, Tocris Bioscience), TAPI-2 (ADAM17 inhibitor, Enzo), GI254023X (ADAM10 inhibitor, Tocris Bioscience), and MMP-2/MMP-9 inhibitor I (MMP-2/MMP-9 inhibitor, Millipore).

Western blotting and protein quantification

Keratinocytes were lysed for 10 min on ice in radioimmunoprecipitation assay lysis buffer (Beyotime) containing a protease inhibitor cocktail (Roche). Eyelid tissue explants were homogenized in the indicated buffer. Lysates were centrifuged at 4°C for 10 min at 14,000 rpm. Protein concentrations were determined using the Bio-Rad protein Assay (Bio-Rad). About 60 μg of lysate was electrophoresed by SDS–polyacrylamide gel electrophoresis, and proteins were blotted onto polyvinylidene fluoride membranes (Millipore). Immunoblots were blocked and probed with antibodies (table S2) as previously described (114). Immunoblots were developed using enhanced chemiluminescence reagents (Bio-Rad). The density of target bands was measured using ImageJ software (National Institutes of Health), and values were normalized against the loading control (GAPDH).

Hematoxylin and eosin staining

Fetuses at E16.5 to E18.5 were harvested from pregnant females that were sacrificed by CO2 asphyxiation. Fetuses were removed from the uterus and dissected in PBS. The limbs were removed for genotype identification by PCR. Subsequently, the fetuses were decapitated, and the heads were fixed, dehydrated, and embedded in paraffin. Five-micrometer sections were mounted onto glass slides and then stained with hematoxylin and eosin. Slides were examined under an Olympus BX53 fluorescent microscope, and images were acquired with an Olympus DP72.

Immunocytochemistry and immunohistochemistry

Wild-type and Spns2ins/ins fetuses at E16.5 to E18.5 were fixed in 4% paraformaldehyde (PFA), processed, and embedded in Tissue Tek optimal cutting temperature compound (Sakura Finetek) for cryosectioning. Fourteen-micrometer eye sections were processed for immunostaining. Keratinocytes were fixed in 4% PFA in PBS for 20 min. For immunohistochemistry, tissue slices and fixed keratinocytes were blocked in 5% normal donkey serum in PBS containing 0.3% Triton X-100 for 1 hour and then incubated with a primary antibody (table S2) overnight at 4°C. After three washes in PBS, the tissue slices and keratinocytes were incubated with the appropriate secondary antibody (Jackson ImmunoResearch) (1:1000) for 2 hours at room temperature, followed by nuclear staining with 4′,6-diamidino-2-phenylindole (Roche). Imaging was performed on an Olympus FV 1200 confocal microscope, and images were acquired with an Olympus DP72.

Generation of the rSpns2 polyclonal antibody

The SPNS2 polypeptide sequence was analyzed, and two epitopes were selected (amino acids 12 to 27, GGAEEEEADAERRRRR-cys, and amino acids 526 to 549, cys-FLSDRAKAEQQVNQLVMPPASVKI). The polypeptides were synthesized by GL Biochem. About 5 mg of each polypeptide was dissolved in 5 ml of PBS, and 0.5 mg of the polypeptide was emulsified with 0.5 ml of Freud’s complete adjuvant (Sigma-Aldrich). Thereafter, 2 ml of blood (pre-immune serum, control) was collected from the ear vein and immediately stored at 4°C to promote clotting. The emulsified mixture was subcutaneously injected into two rabbits at four different sites. The rabbits received immunization boosters every 4 to 6 weeks. Blood was collected at 7 to 10 days after the initial injection, and antibody production was checked by ELISA. The anti-Spns2 serum was stored at −80°C.

Flow cytometry

Peripheral blood was collected from the tail of adult wild-type and Spns2ins/ins rats, and red blood cells were lysed. Peripheral blood mononuclear cells (PBMCs) were suspended in flow cytometry solution containing 2% newborn calf serum, blocked in PBS containing Fc block (0.5 μg/ml; eBioscience), and stained with an anti-rat CD3 fluorescein isothiocyanate, anti-rat CD8a PE, anti-rat CD4 APC, or anti-rat B220 biotin antibody (eBioscience) on ice for 20 min. After washing with PBS, PBMCs were analyzed with the Guava easyCyte HT Flow Cytometer (Millipore).

In situ hybridization

The rSpns2 cDNA was cloned into the pBluescript SK+ vector (Stratagene) and linearized with restriction enzymes. T7 and T3 polymerases were used to synthesize digoxigenin-conjugated antisense and sense rSpns2 riboprobes. The antisense probe was 5′-GGTCCCAGCCACTAAGAGAG-3′, and the sense probe was 5′-GGTCCGAGATAAAGCCAATGAG-3′. A sense probe was used as the control. Wild-type and Spns2ins/ins embryos at various developmental stages were fixed in 4% PFA in PBS overnight. Whole-mount in situ hybridization was performed on 14-μm-thick frozen sections as previously described (10). Sections were incubated with an antidigoxigenin antibody, and the chromogenic reaction was developed with bromochloroindolyl phosphate–nitro blue tetrazolium overnight at 4°C. The slides were viewed under an Olympus BX53 fluorescent microscope.

Quantification of ADAM10 and ADAM17 activity

Endogenous ADAM10 and ADAM17 activity in lysates from wild-type and Spns2ins/ins keratinocytes was measured using the SensoLyte 520 ADAM10 and ADAM17 Activity Assay Kit (AnaSpec) according to the manufacturer’s instructions. The absorbance was read at 450 nm in a microplate spectrophotometer (Infinite 200 PRO, Tecan). Data from at least three independent experiments were analyzed.

Intra-amniotic injection

Wild-type and Spns2ins/ins 16.5-day pregnant rats were anesthetized with 1% pentobarbital and positioned on a heating pad for laparotomy. Liposome 2000 was combined with the vehicle; Spns2 plasmid; the miR-21, miR-221, or miR-222 mimic (GenePharma); or the negative miRNA control, and then injected into the amniotic sac. To investigate the effects of various factors on eyelid closure, saline, EGF (200 ng/ml), S1P (10 μM), FGF10 (100 ng/ml), TGF-β1 (100 ng/ml), or HB-EGF (100 ng/ml) was injected. After 4 days, the status of eyelid closure was examined under an Olympus STZ16 stereomicroscope, and images were acquired with an Olympus MVX10 macroview microscope. At least three embryos were used for each treatment.

Measurement of growth factors in cultured medium and amniotic fluid by ELISA

Medium collected from 48-hour keratinocyte cultures was centrifuged at 1500 rpm for 10 min. The supernatant was obtained, and the concentrations of EGF, HB-EGF, and TGF-α were measured by ELISA (XiTang). The concentration of S1P in amniotic fluid from wild-type and Spns2-deficient embryos at E16.5, E17.5, and E18.5 was also measured by ELISA (Echelon Biosciences).

Statistical analysis

Each experiment presented in the figures is representative of at least three independent experiments. Displayed images and immunoblots are representative of the biological repeats. Data were analyzed with SPSS 20.0 software (SPSS Inc.). In experiments with only two conditions, Student’s t test was used. In experiments with more than two independent groups, one-way analysis of variance (ANOVA) was used, and multiple comparisons were performed when the P value of one-way ANOVA was smaller than 0.05. In the multiple comparisons, the Bonferroni method was used to adjust the P value. All data are presented as means ± SEM. P < 0.05 was considered to be statistically significant.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/11/553/eaat1470/DC1

Fig. S1. Mating type analysis, Spns2 sequencing, and Basic Local Alignment Search Tool (BLAST) results.

Fig. S2. qPCR and in situ hybridization for the mRNA expression of Spns2.

Fig. S3. Alignment of Spns2 from several mammalian species.

Fig. S4. Other phenotypes in Spns2ins/ins rats and specificity of the Spns2 polyclonal antibody.

Fig. S5. S1P concentrations in keratinocyte supernatants and cell lysates.

Fig. S6. The number of stress fibers within keratinocytes.

Fig. S7. Quantification of other proteins within the EGFR signaling pathway.

Fig. S8. MMP-2/MMP-9 inhibitor I does not affect S1PR-mediated transactivation of EGFR signaling.

Fig. S9. ADAM10 and ADAM17 expression in Spns2-deficient keratinocytes.

Fig. S10. Abundance of cFos, Mig6, and proteins in the RhoA pathway in wild-type and Spns2ins/ins keratinocytes.

Table S1. Primers used in this study.

Table S2. Antibodies used in this study.

Movie S1. Absence of blinking reflex in mutant rats.

REFERENCES AND NOTES

Acknowledgments: We thank J. Qiu and L. Fei for technical assistance and Z. Tan for statistical suggestions. Funding: This study was supported by the National Natural Science Foundation of China (31571428, 31271127, 81371364, and 81401019) and Hebei province science and technology research project (ZD2016094). Author contributions: J.W. and G.J. designed the experiments and wrote the paper; G.B., C.Y., L.L., and C.F. carried out most of the experiments and image analysis; K.C., P.R., Q.Z., F.L., K.Z., Q.X., J.X., J.S., and Y.Z. performed the experiments. W.W., S.K.C., and R.S. provided valuable advice on the research. All authors discussed the results and the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: WGS data have been deposited into the SRA of the NCBI (www.ncbi.nlm.nih.gov/sra) with the accession number SRR7415781. All other data needed to evaluate the conclusions in this study are available in the paper or the Supplementary Materials.
View Abstract

Navigate This Article