Loss of Sphingosine Kinase 1/S1P Signaling Impairs Cell Growth and Survival of Neurons and Progenitor Cells in the Developing Sensory Ganglia

Background Lysophospholipids such as lysophosphatidic acid (LPA) and sphingosine-1-phosphate (S1P) are important signaling molecules that can regulate a wide range of cellular responses. We discovered that Sphingosine kinase 1 (Sphk1), a key enzyme that converts sphingosine to S1P, is expressed in neurons and progenitor cells in nascent trigeminal and dorsal root ganglia during mouse embryogenesis. Methods and Findings Sphk1 null mouse embryos do not display overt deficits owing to compensation by Sphk2. Thus, we analyzed embryos that are deficient in both Sphk1 and Sphk2 (which essentially eliminates S1P function) in order to investigate the role(s) of Sphk1 during sensory ganglia formation. While animals lacking 1–3 alleles of Sphk1 and Sphk2 had no obvious phenotype, embryos without both genes displayed clear developmental defects. The complete absence of Sphk1 and Sphk2 resulted in trigeminal and dorsal root ganglia with fewer neurons and progenitor cells. The profound loss in cell number could be attributed to a decrease in cell proliferation as well as an increase in apoptosis. Furthermore, Sphk1/2 double mutants displayed an overall reduction in other sphingolipids as well as an imbalance of S1P/sphingosine and S1P/ceramide ratio, thereby favoring cell death and reducing cell growth. Conclusions Together, these results provide strong in vivo evidence that sphingosine kinase/S1P signaling plays an important role in regulating early events during development of sensory ganglia.


Introduction
Sensory neurons convey senses, among others, of touch, temperature, and proprioception from the periphery to the central nervous system. The sensory system is made up of the cranial ganglia (which include the trigeminal ganglia and dorsal root ganglia). Both neurons and glia in the dorsal root ganglion are derived from the multipotent neural crest cells and the trigeminal ganglion receives additional neuronal contribution from the trigeminal placode in the head [1]. As sensory neurons differentiate and innervate target sites, their survival is supported by neurotrophic factors such as the family of neurotrophins [2,3,4,5]. On the other hand, what regulate progenitor cell proliferation and neuronal survival during early ganglia development is less well understood. Ciliary neurotrophic factor and neurotrophin-3 have been demonstrated to induce proliferation of progenitor cells and neuronal differentiation in immature dorsal root ganglia [6]. We recently discovered that Sphingosine kinase 1 was expressed in progenitor cells and a subset of neurons in the nascent sensory ganglia and hypothesized that the Sphingosine kinase/sphingosine-1-phosphate could mediate cell proliferation and survival prior to target innervation.
Lysophospholipids such as lysophosphatidic acid (LPA) and sphingosine-1-phosphate (S1P) are membrane-derived bioactive lipid mediators [7,8,9]. S1P, its precursor, sphingosine, and ceramide are sphingolipids that have been studied by many groups as they are involved in signaling pathways that control multiple cellular processes including cell proliferation, differentiation, survival and migration ( [8,9,10,11]. Sphingosine is phosphorylated by Sphingosine kinases to form S1P [12] and sphingosine can be generated either by degradation of sphingolipids or by deacylation of ceramide. S1P can be dephosphorylated by Sphingosine-1-phosphate phosphatases 1 and 2 (Sgpp1 and Sgpp2) or by lipid phosphohydrolases with broader substrate specificity. Alternatively, S1P can be irreversibly degraded to ethanolamine phosphate and hexadecenal (long chain aldehyde) by the action of sphingosine-1-phosphate lyase.
Sphingolipids regulate cell growth and survival in many cell types. S1P is usually associated with cell proliferation and survival; in contrast, ceramide and sphingosine are usually associated with cell death [11,12,13,14]. Stressful stimuli or growth factor withdrawal can activate sphingomyelinase, which converts sphingomyelin to produce ceramide. Ceramide can take part in multiple events that lead to stress response, growth arrest, or apoptosis. The enzyme ceramidase can further metabolize ceramide to produce sphingosine, which can inhibit protein kinase C and induce apoptosis. Thus, the relative levels of S1P, ceramide, and sphingosine inside a cell can influence survival or apoptosis [13,14,15]. These sphingolipid molecules are interconvertible, higher level of S1P favors cell growth, whereas cell death ensues when the reaction tends towards sphingosine/ceramide; thus, this balance is sometimes referred to the ''sphingolipid rheostat'' [16,17,18]. Of all the molecules that participate in the sphingolipid rheostat, Sphingosine kinase is particularly important because it can decrease ceramide and sphingosine by catalyzing the formation of S1P. In mammals, two sphingosine kinases, Sphingosine kinase 1 (Sphk1) and Sphingosine kinsae 2 (Sphk2) have been identified. The biological function of Sphk2 is controversial and less well characterized than Sphk1. Sphk1 is upregulated in several types of cancers, including cancer of the colon, breast, stomach, and kidney [18]. Additional evidence also suggested that tumor cells can overexpress Sphk1, implicating the S1P pathway in the uncontrolled growth and dampening of cell death associated with cancers [18].
Many actions of S1P are mediated by G protein coupled receptors. In mammals there are at least five S1P receptors (S1pr1-5) [19]. Some of the S1P receptors are expressed in the nervous system [20,21] and play a role in the development of neurons and glia cells. S1pr1, for example, is necessary for brain development [22] and the survival of oligodendrocyte precursors requires the activity of S1pr5 [23]. However, S1P can also act as a second messenger inside cells and elicit various cellular responses [24,25]. In addition to mobilizing intracellular Ca 2+ [24], recent studies reported that intracellular S1P could regulate histone deacetylases and the ubiquitin ligase activity of tumor necrosis factor receptorassociated factor 2 [10,26,27]. Here, we demonstrate a role for the S1P pathway in sensory ganglion development and show that complete loss of Sphk activity leads to severe disruption of neuronal survival and proliferation of progenitor cells in the developing trigeminal and dorsal root ganglia.

Ethics statement
All studies using mice were conducted according to Institutional Animal Care and Use Committee (IACUC) guidelines of the Medical College of Wisconsin with approval, protocol AUA198. Mice were housed and maintained in the Bioresource Center of the Medical College of Wisconsin, which is accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care (AAALAC).
Whole mount in situ hybridizations E9.5 (22-25 somites) and E10.5 (32-35 somites) wild type, Sphk12/2; Sphk2+/2 Sphk12/2; Sphk22/2 embryos were fixed in 4% paraformaldehyde overnight at 4uC, rinsed in phosphate buffered saline, and dehydrated in methanol series and stored in 100% methanol at 220uC. Embryos were processed for whole mount in situ hybridization as previously described [31] and color reaction developed with BM Purple (Roche). Whole mount and section in situ hybridizations for each probe were repeated at least three times to confirm our results.

Section in situ hybridization and immunfluorescence
For section in situ hybridizations, E10.5 (32-35 somites) and E11.5 (40-42 somites) embryos were fixed with modified Carnoy's solution, processed for paraffin sectioning (10 mm), and in situ hybridization as described [32]. For post-in situ immunohistochemistry, slides were incubated in Tuj-1 or SOX10 overnight at 4uC [33]. Fluorochrome conjugated secondary antibodies (Jackson ImmunoResearch) were applied for 2 hours at room temperature in dark. Brightfield in situ hybridization and fluorescence immunostaining signals on sections were photographed using a Zeiss Z1 microscope equipped with a monochrome MRm AxioCam camera. Brightfield signals were inverted in Adobe Photoshop (from purple brightfield signal to green) and overlayed with Tuj-1 and SOX10 fluorescence signals. For immunofluorescence, wild type, Sphk12/2; Sphk2+/2 Sphk12/2; Sphk22/2 embryos were embedded in gelatin, frozen in liquid nitrogen, and stored in 280uC. Serial cryosections (10 mm) were cut from head and trunk (between forelimb and hindlimb) regions of embryos and adjacent sections were collected on two sets of slides. One set of slides was stained for islet-1/2 and active-caspase 3 (act-Casp-3) and the second set of slides was stained for SOX10 and phosphohistone H3 (PH3).

Cell counting and statistics
All sections including trigeminal ganglia and dorsal root ganglia (the first three DRG at the fore limb level) were photographed and Tuj-1+, active-caspase 3+ (from one set of slides) and SOX10+, phospho-histone H3+ (a second set of slides with alternate sections) cells were counted using ImageJ. One trigeminal ganglion and two DRG from each embryo were counted, and three embryos for each group were used in the analyses. Data are expressed as means 6 standard error of means. Student's t-test was used for statistics, p values of p,0.01 or p,0.05 were considered significant.
Measurement of S1P, sphingosine and ceramide levels Three E11.5 (40-42 somites) embryos from wild type, Sphk12/2; Sphk2+/2, and Sphk12/2; Sphk22/2 group were collected and frozen immediately at 280uC. S1P, sphingosine, and ceramide analyses were performed by the Lipidomics core of Medical University of South Carolina. Data are expressed as means 6 standard error of means. Student's t-test was used for statistics, p values of p,0.01 or p,0.05 were considered significant.

Sphingosine kinase 1 is expressed in the developing trigeminal and dorsal root ganglia
We have previously shown that the developing sensory ganglia express very low levels of S1P receptors during early gangliogenesis ( [20] and data not shown). To investigate the potential role of another key component in the S1P pathway, Sphingosine kinase 1 (Sphk1), in the developing sensory ganglia, we first examined its expression in these structures. We performed section in situ hybridization on trigeminal and dorsal root ganglia sections in E10.5 (32-35 somites) and E11.5 (40-42 somites) embryos.
At E10.5, Sphk1 transcript was scattered throughout the trigeminal ganglion but was not expressed by all the cells (Fig. 1A, B). To identify which cell types expressed Sphk1, we colabeled the same section with Tuj-1 and SOX10 to identify neurons and neural crest progenitor cells, respectively. By examining higher magnification of single labeled and merged images, we found that Sphk1 was expressed in many SOX10+ cells but not Tuj-1+ cells at this stage ( Fig. 1C-F). At E11.5, Sphk1 expression pattern appeared similar to E10.5. Co-labeling experiments showed that Sphk1 continued to be observed in SOX10+ cells but now some neurons also began to express the Sphk1 mRNA at E11.5 ( Fig. 1G-L).
At E10.5, we detected Sphk1 in the periphery of the dorsal root ganglia. Immunostaining with SOX10 in the same section confirmed that Sphk1 co-localized with some SOX10+ cells, suggesting that progenitor cells expressed Sphk1. We did not detect Sphk1 transcripts in DRG neurons (Tuj-1+) at this stage ( Fig. 2A-F). At E11.5, we observed Sphk1 in the periphery of the DRG and some scattered Sphk1+ cells were present throughout the ganglion. Co-labeling experiments showed that Sphk1 was expressed in both SOX10+ progenitors as well as Tuj-1+ neurons at this time ( Fig. 2G-L).

Depleting Sphingosine kinase activity causes defects in sensory ganglia
To investigate the possible roles of Sphk1 in the developing nervous system, we aimed to ablate Sphk1 in the mouse embryo. In mammals, there are two Sphingosine kinases, Sphk1 and Sphk2, and knockout mice for both have been generated [20,22]. Sphk12/2 or Sphk22/2 single mutants do not show any obvious phenotype, survive until adulthood, and are fertile [22,28]. Previous studies demonstrated that Sphk2 mRNA levels were not altered in Sphk1 null animals compare to controls. Furthermore, S1P levels were found to be similar or just slightly decreased in Sphk12/2 tissues compared to wild type [28]. As we do not observe any sensory ganglia defects in the Sphk1 single mutant (Fig.  S2), Sphk2 may be able to compensate for the absence of Sphk1 by maintaining sufficient S1P albeit it is expressed at low levels in sensory ganglia (Fig. S1).
In contrast to Sphk1 or Sphk2 single mutant, Sphk1/2 double knockout mice have severe phenotypes [22,28]. Deleting both Sphk1 and Sphk2 eliminates S1P function, and because S1P is important for vascular development, Sphk12/2;Sphk22/2 embryos die at E12.5 due to vascular defects [22]. We generated Sphk12/2;Sphk22/2 double mutant embryos in order to examine the role of sphingosine kinase activity in sensory ganglia development. Similar published results, we observed that Sphk12/2;Sphk22/2 double mutant mice die at E12-12.5 and the embryos are slightly smaller than their littermates and wild type embryos [22]. We examined trigeminal and dorsal root ganglia in Sphk12/2;Sphk22/2 double mutant embryos at E11.5 (40-42 somites) and compared them to controls. In wild type and Sphk12/2;Sphk2+/2 animals, trigeminal and dorsal root ganglia were formed properly and the morphology of neurons was normal, as demonstrated by Tuj-1 immuostaining. In contrast, Sphk12/2;Sphk22/2 sensory ganglia appeared smaller and disorganized ( Fig. 3A-F). There appeared to be fewer Tuj-1+ neurons in both trigeminal and dorsal root ganglia in Sphk12/2;Sphk22/2 embryos. These defects will be analyzed in more details and presented in the following sections of the manuscript. Neural crest stem cells migrate and form trigeminal and dorsal root ganglia [1]. To dissect the basis of sensory ganglia dysgenesis in Sphk12/2;Sphk22/2 animals, we first compared the migration pattern of neural crest cells in control and Sphk12/2;Sphk22/2 embryos using Sox10 whole mount in situ hybridization. E9.5 (22-25 somitets) and E10.5 (32-35 somites) wild type, Sphk12/2;Sphk2+/2 and Sphk12/2;Sphk22/2 embryos were processed and compared together. At E9.5, the forming trigeminal ganglion and hindbrain neural crest cells can be visualized in the head. In the trunk, neural crest cells are beginning to migrate at the level of the first few somites. Migrating neural crest cells and trigeminal ganglia could be observed in E9.5 embryo in all three genotypes ( Fig. 4A-C). Although the Sphk12/2;Sphk22/2 embryos were slightly smaller, both cranial and trunk neural crest cells showed normal migratory patterns. By E10.5, the trigeminal ganglion is clearly formed and most of the trunk neural crest cells are migrating. The pattern of migration of neural crest cells did not seem to be altered in Sphk12/2; Sphk22/2; however, the streams of migrating cells appeared to be thinner compare to controls ( Fig. 4D-F). Thus, neural crest formation and migration occurred in Sphk12/2;Sphk22/2 mutant embryos and the defects observed in the sensory ganglia might take place shortly after their formation.

Ablation of Sphingosine kinases disrupted trigeminal ganglion development
Because Sphk12/2;Sphk22/2 trigeminal ganglia appeared smaller and might contain fewer cells at E10.5 ( Fig. 3D-F), we carried out a series of double immunostaining experiments at this embryonic stage and quantitatively studied different cell populations, together with cell death, and cell proliferation assays. First, we assessed cell death/apoptosis. We used anti-activated caspase 3+ (act-Casp3) and anti-Islet-1/2 antibodies to determine the number of dying cells and neurons, respectively. In wild type ganglia, there were a small number of act-Casp3+ cells as expected due to naturally occurring cell death at this age. When we compared the number of dying cells in wild type, Sphk12/2;Sphk2+/2, and Sphk12/2;Sphk22/2 trigeminal ganglia, there was a slight increase in cell death in the Sphk12/2;Sphk2+/2 and double mutants ( Fig. 5A-C, H, Table 1). Next, we assessed cell proliferation. We examined the number of mitotic cells and progenitor cells by using phosphohistone H3 (PH3) and SOX10 as markers. We found that there were 334618 dividing cells in wild type and 33169 in Sphk12/2;Sphk2+/2, while Sphk12/2;Sphk22/2 trigeminal ganglia only contained 263621 PH3+ cells (Fig. 5D-F). Some of the PH3+ cells also co-labeled with SOX10 ( Fig. 5D, E). Although the number of PH3+ cells was lower (263621) in Sphk12/2;Sphk22/2 compared to controls (,331), the percentage of dividing cells compared to the total number of cells in the ganglia was essentially the same amongst these genotypes (,2.5% on average, Table 1).
To determine if depleting Sphingosine kinases affected both neurons and progenitor cells, we counted the total number of islet-1/2+ neurons and SOX10+ progenitors cells in the trigeminal ganglia at E10.5. In wild type and Sphk12/2;Sphk2+/2 trigeminal ganglia, there were 6,018 and 6,302 Islet-1/2+ neurons, respectively;  Table 1). We also observed a significant decrease (26%) in the progenitor population. Sphk1/2 double mutant ganglia had only 5,373 SOX10+ cells compared to 7,252 in wild type and 7,245 in Sphk12/2;Sphk2+/2. However, because neither cell proliferation nor cell death was greatly affected in the double mutant trigeminal ganglia at E10.5, it is possible that the reduced number of neurons and progenitors was caused by a decrease in neuronal differentiation and/or a scarcity of progenitor cells.
Survival of trigeminal ganglion cells was compromised by the loss of Sphk activity as an increase in cell death was observed at E11.5. We performed cell death and proliferation analyses on E11.5 trigeminal ganglia and quantified cell numbers. In wild type and Sphk12/2;Sphk2+/2 trigeminal ganglia, only 0.3-0.8% (61-152 act-casp3+ cells out of 18,899-19,755 total cells) of all cells were dying; on the other hand, there were significantly more act-casp3+ cells throughout the Sphk12/2;Sphk22/2 double mutant ganglia (23436369, 19.5% of total cells; Table 1, Fig. 6A-C, H). In addition, the number of dividing cells was greatly reduced in Sphk12/2;Sphk22/2 compared to control embryos ( Fig. 6D-F,  H). In Sphk1/Sphk2 double mutants, there were only 1764 PH3+ cells compared to 397611 in wild type and 402634 in Sphk12/2;Sphk2+/2 trigeminal ganglia. When we counted the total number of neurons and progenitor cells, we found that wild type and Sphk12/2;Sphk2+/2 trigeminal ganglia had an average of 10,000 neurons whereas Sphk12/2;Sphk22/2 had about 5,531 ( Table 1). The number of SOX10+ progenitors was also significantly lower. While wild type and Sphk12/2;Sphk2+/2 ganglia had an average of 9,500 SOX10+ cells, the double mutant only had ,6,463 such cells. Thus, both neurons and non-neuronal cells showed a 35-40% loss in cell number when Sphk1 and Sphk2 were deleted. Survival of both neurons and progenitor cells were affected by the absence of Sphks as act-casp3+ cells were found in both islet-1/2+ and SOX10+ populations (Fig. S3).

Deleting Sphks negatively influenced dorsal root ganglion development
Next we examined if the spinal sensory ganglia defects were similar to those observed in the trigeminal ganglia. At E10.5, there were few act-Casp3+ cells (Fig. 7A-C, H) and only several PH3+ cells (Fig. 7D-F, H) in dorsal root ganglia from all genotypes. Although the number of SOX10+ cells in Sphk12/2;Sphk22/2 was slightly lower compared to wild type and Sphk12/2; Sphk2+/2 dorsal root ganglia, the difference was not significant. Some PH3+ were dually positive for SOX10 (Fig. 7D-F, G). Even without major changes in cell death and proliferation, the double mutant DRG were still significantly smaller in size and there appeared to be fewer islet-1/2+ cells. Indeed when we counted the neurons, we found that there were ,328 islet-1/2+ neurons in the   Table 1). Thus, the double mutant DRG had less than half the number of neurons compare to controls, suggesting that the lack of Sphk activity negatively influenced neuronal differentiation or survival or both of these events during the initial formation of the DRG.
By E11.5, cell death and proliferation were affected in Sphk12/2;Sphk22/2 mutant DRG. At this stage, there were only a small number of dying cells in wild type and Sphk12/2; Sphk2+/2 ganglia, as normal cell death has not commenced [34].

Deficits in sensory ganglia development can be attributed to the loss of Sphingosine kinase activity
Many effects of the S1P pathway are mediated through S1P receptors. In the Sphk12/2;Sphk22/2 embryos, cell death and proliferation were affected in the brain at E11.5. The developing brain expresses abundant S1pr1 [20,22], and cell death and proliferation defects were obvious at E11.5 in S1pr1 null animals [20]. These data suggested that perhaps S1P, acting through S1pr1, was at least partially responsible for CNS development. Besides the receptor-mediated activities of S1P, our data support a role for the intracellular activity of SPHK in regulating sensory ganglia development. To begin to address this issue, we compared trigeminal and dorsal root ganglia from S1pr1 and S1pr2 mutant mice [29,30,35] and Sphk1;Sphk2 double knockout embryos. While S1pr22/2 mouse do not display overt neural phenotypes until three to seven weeks of age [30], S1pr12/2 embryos die between E12.5-E14.5 due to embryonic hemorrhage [29]. In contrast to Sphk1/2 double mutants trigeminal and dorsal ganglia from E11.5, which were much reduced in size and cell number, S1pr1 and S1pr2 null sensory ganglia were similar to their control littermates ( Fig. 9 and data not shown).
To investigate if the effects of ablating Sphks were specific to sensory ganglia, we examined the sympathetic ganglia, which are also neural crest derived. At E11.5, Sphk1 transcript was not detected in the sympathetic ganglia (Fig. 10). When we compared wild type, Sphk12/2;Sphk2+/2, and Sphk12/2; Sphk22/2 embryos, sympathetic ganglia appeared normal in all three genotypes (Fig. 10). This suggests that deleting Sphks did not impose non-specific defects in all periphery ganglia.
When we examined the data in more detail, we found that the sphingolipid levels were decreased across the board and S1P was absent in the double mutant. The balance of ceramide/S1P has been proposed as a ''rheostat'' to regulate cell growth and cell death [17,18,36]. Thus, we compared the relative ratios of ceramide/S1P and Sph/S1P in control and Sphk12/2;Sphk22/2 animals. In wild type, the ratio of ceramide/S1P ratio was 28.2 and Sph/S1P was 3.4. In Sphk12/2;Sphk2+/2 embryos, ceramide/S1P and Sph/S1P ratios were elevated to 78.5 and 13.6, respectively. Because there was almost no S1P in the Sphk12/2;Sphk22/2 double mutant, the ceramide to S1P ratio sharply increased to 2,263.70 and Sph/S1P to 377.9 (Table 2).

Discussion
We have investigated the role of Sphk1, a key component in S1P signaling, in the developing sensory ganglia. We found that Sphk1 was expressed first in neural crest progenitors then neurons in newly formed trigeminal and dorsal root ganglia. To investigate its function during development, we generated Sphk1/Sphk2 double mutant mice that completely lacked Sphk activity. Absence of Sphks leads to massive cell death in neurons and progenitors as well as near complete suppression of cell proliferation in the progenitor population. The Sphk1 and Sphk2 double knockout results showed that the phenotypes in the trigeminal and dorsal root ganglia are robust at E11.5-both neurons and progenitor cells are greatly depleted due to a combination of an increase in cell death and a decrease in cell proliferation. Upon examination of sphingolipid levels in the Sphk mutant, we saw an overall decline in multiple sphingolipids and essentially no S1P. Our data provide strong in vivo evidence that sphingosine kinase/S1P signaling has a novel role in regulating cell survival during development of sensory ganglia. Establishment of the sensory ganglia involves proper migration and condensation of neural crest stem cells, differentiation, and cell proliferation. In the Sphk1/2 double mutant, we did not observe overt disruption of neural crest migration pattern, Figure 7. Ablation of Sphk1 and Sphk2 affects neuron but not progenitor cell numbers in E10.5 dorsal root ganglion. (A-C) Transverse trunk sections from wild type, Sphk12/2;Sphk2+/2, and Sphk12/2;Sphk22/2 immunostained with islet-1/2 (green, nuclear) and act-Casp-3 (red). Wild type (A) and Sphk12/2;Sphk2+/2 (B) DRG appear similar in shape and size. Sphk12/2;Sphk22/2 DRG (C) is smaller and has fewer neurons. Few act-Casp-3+ cells are observed in DRG from all three genotypes. (D-F) Adjacent sections to those used in (A-C) were processed for PH3 (red) and SOX10 (green, nuclear) immunofluorescence. The shapes of the DRG are similar in wild type (D) and Sphk12/2;Sphk2+/2 (E); most SOX10+ cells are located in the periphery of the ganglion. Sox10+ cells in Sphk12/2;Sphk22/2 DRG are no longer restricted to the periphery of the ganglion. PH3+ (green) cells can be found in DRG from all genotypes examined and some of them are also SOX10 positive (arrow). (G) Quantification of cell number in E10.5 DRG demonstrates that there are 24% fewer SOX10+ cells in Sphk12/2;Sphk22/2 DRG compared to wild type and Sphk12/2;Sphk2+/2 but this reduction is not statistically significant (p = 0.079). The double mutant DRG shows a 56% reduction in the number of neurons compared to wild type and Sphk12/2;Sphk2+/2 embryos. n = 3 embryos from each genotype; ** = Sphk12/2;Sphk22/2 to wild type comparison, p ,0.01. {{ = Sphk12/2;Sphk22/2 to Sphk12/2;Sphk2+/2 comparison, p,0.01. (H) Quantification of total number of dying (act-Casp-3+) and dividing (PH3+) cells. Cell count results reveal no significant difference in cell death and proliferation in Sphk12/2;Sphk22/2 DRG compared to wild type and Sphk12/2;Sphk2+/2 at this stage. NT = neural tube. Scalebar for A-F = 50 mm. doi:10.1371/journal.pone.0027150.g007 although the streams of cells appeared to be thinner. This suggests that depleting S1P in the embryo does not affect neural crest cell formation and migration per se but may affect their survival and/ or proliferation.
Although the actions of S1P are usually transmitted through binding to its receptors, the sensory ganglia phenotype observed here could be explained at least in part, by SPHK activities. In contrast to the massive defects we observed in Sphk12/2;Sphk22/  . Trigeminal ganglia develop normally in S1pr1 and S1pr2 mutant mice. E11.5 (40-42 somites) trigeminal ganglia were sectioned processed for Tuj-1 immunostaining (brown), hematoxylin was used as a counterstain (blue). (A-C) Trigeminal ganglia from control, S1pr12/2 and S1pr22/2 appeared similar in size and contain many Tuj-1+ neurons. (D) A section from Sphk12/2;Sphk22/2 trigeminal ganglion was included here for comparison. It was much smaller compared to S1pr12/2 and S1pr22/2 trigeminal ganglia. Scalebar for A-D = 100 mm. doi:10.1371/journal.pone.0027150.g009 2, sensory ganglia from S1pr1 and S1pr2 mutant embryos are largely similar to their heterozygous litters and wild type controls. These data suggest that Sphk1 and intracellular S1P signaling can play a role in sensory ganglia development.
Ceramide and sphingosine regulate stress response pathways and usually induced cell death and growth arrest. On the other hand, S1P tends to promote cell proliferation and inhibit cell death. Thus, the balance of intracellular ceramide, sphingosine, and S1P is important in regulating cell growth and survival [17,18,36]. In the Sphk1/2 double mutant, there was essentially no S1P. Hence, we hypothesized that the massive decrease in cell survival of sensory ganglia lacking Sphks was caused by a large increase in ceramide and sphingosine in the absence of S1P. When we measured the levels of these sphingolipids, we found that there was no S1P in the double mutant as expected. There was also a four-fold decrease in S1P in Sphk12/2;Sphk2+/2 embryos compared to wild type, which is also expected. Ceramide level was decreased by 2-fold in Sphk12/2;Sphk22/2 double knockout and there was a slight decrease in sphingosine. This was a bit surprising as we expected a buildup of sphingosine due to the absence of SPHK activity. We then considered the balance of these sphingolipids and sought to examine if they were altered. The ratio of ceramide:S1P was 28.2 in wild type and 78.5 in Sphk12/2;Sphk2+/2 embryos; however, this ratio rose dramatically to 2,263.7 in the double mutant due to the absence of S1P. The sphingosine/S1P ratio was 3.4 in wild type and 377.8 in Sphk12/2;Sphk22/2 mutants, over a 100-fold increase. Therefore, although the pro-apoptotic ceramide and sphingosine were not increased, the lack of S1P disrupted the sphingolipid balance inside cells, favoring cell death and suppressing cell proliferation and survival. Our current study demonstrates that the relative ratios of these molecules are important in regulating cell survival in vivo.
During sensory ganglia development, several signaling pathways have been implicated in the survival and maintenance of neurons and precursor cells. For example, the family of neurotrophins and their receptors (Trk A, B, C) are involved in sensory neuron survival in both trigeminal and dorsal root ganglia [2,5,37]. It is unlikely that Sphk1 is directly downstream of Trk receptors because it was clearly demonstrated that only neurons but not progenitor cells express Trk receptors in murine sensory ganglia [2] whereas Sphk1 is expressed in both cell populations. It is possible that the p75 receptor pathway may indirectly influence Sphk1 activity as it can activate ceramide and apoptotic pathways [38,39,40]. Sphk1 is expressed by E10.5 in the sensory ganglia when p75 receptor can be observed; thus, it is plausible that Sphk1 can act downstream of  the p75 receptor directly or indirectly through ceramide. In addition to regulating cell death and proliferation in the developing nervous system, S1P/Sphingosine kinase signaling may modulate neuronal survival and functions in adult animals. For instance, in an in vitro model of brain ischemia, S1P was shown to reduce cell death, possibly via protein kinase C e translocation to the mitochondria as well as reduction of Ca 2+ in the mitochondria [41]. In adult hippocampal CA3 neurons, synaptic localization and activation of S1PR3, and generation of long-term potentiation were dependent on the production of S1P by SPHK1, as these activities were completely abolished in Sphk1 null animals [42]. These data suggests that S1P/Sphingosine kinase signaling exert wide-ranging effects in the nervous system, some of which are receptor independent while others may require interactions with S1P receptors.
In summary, we have shown that the complete absence of Sphingosine kinase/S1P activity has a detrimental effect on early sensory gangliogenesis-assigning a novel role for sphingosine kinase in the developing nervous system. Our future goal is to identify signaling pathways that interact with S1P and mediate the proper formation of sensory ganglia. Figure S1 Sensory ganglia in Sphk12/2 and Sphk22/2 single knockout mice do not display any obvious defects. Head and trunk transverse sections from E11.5 (40-42 somites) wild type, Sphk12/2 and Sphk22/2 mice were processed for Tuj1 (red) immunofluorescence and counterstained with DAPI (blue) to examine morphologies of the trigeminal (Trig) and dorsal root ganglia (DRG). (A,B) Trunk sections showing E11.5 DRG in Sphk12/2 and Sphk22/2 embryos. DRG from Sphk12/2 (A) and Sphk22/2 (B) are similar in shape, size, and contain many Tuj1+ neurons; they look similar to wild type and Sphk12/2;Sphk2+/2 DRG in Fig. 3A Figure 11. Schematic of S1P-SPHK1 and model of S1P-Sphingosine/ceramide balance. (A) A highly simplified schematic of S1P metabolic pathway. Sphingosine (Sph) is catalyzed by Sphingosine kinase 1 or 2 (Sphks) to form S1P. S1P can be degraded reversibly by Sphingosine phosphatase 1 or 2 (Sgpps) back to sphingosine or irreversibly by Sphingosine-1-phosphate lyase (Sgpl) into ethanolamine phosphate and hexadecenal. Sphingosine can be converted to ceramide (Cer) by Ceramide synthase, the reverse reaction is catalyzed by Ceramidase. Cumulative evidence from the literature suggests that S1P favors cell proliferation and promote survival while ceramide and sphingosine tend to drive cells into apoptosis and cell cycle arrest. (B) In wild type embryo where there is equilibrium of sphingosine/ceramide and S1P, cell death and survival are kept in balance. In the Sphk12/2;Sphk22/2 double mutant, not only is there an overall decrease in sphingolipids including sphingosine and ceramide, S1P is completely absent. This results in a distortion of the sphingosine/ceramide to S1P ratio in the cells, which can tip the scale towards cell death and decrease in cell proliferation. doi:10.1371/journal.pone.0027150.g011