Lack of lipid phosphate phosphatase-3 in embryonic stem cells compromises neuronal differentiation and neurite outgrowth

Abstract

Bioactive lipids such as lysophosphatidic acid (LPA) and sphingosine-1-phosphate (S1P) have been recently described as important regulators of pluripotency and differentiation of ES cells and neural progenitors. Here we show that LPP3, an enzyme that regulates the levels and biological activities of the aforementioned lipids, participates in neural differentiation and neuritogenesis. We find that Lpp3^−/− ES cells differentiated in vitro into spinal neurons, show a considerable reduction in the amount of neural precursor and young neurons formed. In addition, differentiated Lpp3^−/− neurons exhibit impaired neurite outgrowth. Surprisingly, when Lpp3^−/− ES cells were differentiated, an unexpected appearance of smooth muscle actin positive cells was observed, an event that was dependent upon phosphorylated sphingosines. Our data suggest that LPP3 plays a fundamental role during early spinal neuroepithelium development and that it could also be instrumental in regulating neurite and axon outgrowth in vivo.

Introduction

Embryonic stem (ES) cell research is a powerful tool for understanding the mechanisms of pluripotency, as well as the cellular mechanisms and signaling pathways that govern differentiation to particular cell lineages. The role of bioactive lipids in stem cell biology is a relatively novel topic in this field and only recently their crucial roles are starting to be revealed. For example, sphingosine-1-phosphate (S1P) and lysophosphatidic acid (LPA) are now known to act as positive regulators of mouse ES cell (mES) proliferation as well as human ES (hES) cell long-term self-renewal (Pitson and Pebay, 2009). However, these lipids also play important roles in ES differentiation: S1P has a potent cardiomyocyte-inducing activity in differentiated mES (Sachinidis et al., 2003) and administration of LPA in high concentrations, during neural differentiation of hES, blocks neurosphere and young neuron formation (Dottori et al., 2008). Although the roles of lysophospholipid G-protein coupled receptors (S1P_1-5, LPA_1-5) are known for a wide variety of cellular processes including cell proliferation, apoptosis, neurite retraction, axon guidance and cell survival (Kingsbury et al., 2003; Toman et al., 2004; Fukushima and Morita, 2006; Strochlic et al., 2008; Yamazaki et al., 2008; Yamane et al., 2010), little is understood about how these lysophospholipids regulate cell differentiation.

It is, however, currently known that bioactive lipids play an extensive and fundamental role in early nervous system development. Double knockout embryos for sphingosine kinase 1 and sphingosine kinase 2 (Sphk1/2), display clear defects in neural tube closure and increased apoptosis in the nervous system (Mizugishi et al., 2005). In this context, multiple publications have illustrated that the balance between S1P and its precursor ceramide is critical for inducing either cell survival or death in various neural cell types (Edsall et al., 1997; Hartfield et al., 1997; Shinpo et al., 1999). Furthermore, in Xenopus laevis sphingosine kinase-1 expression limits the pathway by which retinal axons grow from the optic chiasm to the tectum and either gain or loss of S1P function in vivo causes errors in axon navigation (Strochlic et al., 2008). It has also been shown that autotaxin (Atx), the main extracellular-LPA synthesizing enzyme, contributes to anterior brain morphogenesis and establishment of the midbrain-hindbrain boundary (MHB) (Ohuchi et al., 2010; Koike et al., 2011). Another report indicates that in a neurosphere culture system, 1 μM of LPA increases the number of MAP2-positive neurons and decreases the proportion of myelin basic protein-positive oligodendrocytes in cortical neuroblasts (Fukushima et al., 2007). It has also been suggested that low concentrations of LPA enhance proliferation, increase the percentage of MAP2- and ChAT-immunoreactive neurons and positively regulate cell migration and neurite extension in rat embryonic neural stem cell culture (Cui and Qiao, 2006). Moreover, LPA is also capable of inducing cortical growth and folding in the mouse brain (Kingsbury et al., 2003) and recently it has been implicated in the induction of neuronal polarity (Yamane et al., 2010).

Lipid phosphate phosphatases (LPPs) are a group of integral membrane enzymes with a broad lipid phosphate phosphohydrolase activity. S1P and LPA are two of the four known substrates of these enzymes. Dephosphorylation of these lysophopholipids by LPPs attenuates their signaling activity (Sigal et al., 2005; Brindley and Pilquil, 2009). Conversely, inactivation of these enzymes could promote either enhanced activity (Smyth et al., 2003) or downregulation of their receptors (Lopez-Juarez et al., 2011). In this context, studying the effects of altered LPP expression could contribute to understanding the roles of these lysophospholipids in a variety of cellular processes. Of the three described mammalian LPPs, LPP3 seems to have a specific and non-redundant role during development. The latter is supported by the lethality that Lpp3 KO animals display at early stages of development (Escalante-Alcalde et al., 2003), while a gene trap allele of Lpp1 and Lpp2-knockout mice are viable (Zhang et al., 2000; Tomsig et al., 2009). In addition, it has been described that these enzymes have different subcellular localizations, suggesting distinct functions in different cellular compartments (Jia et al., 2003; Kai et al., 2006). In the developing nervous system, LPP3 is expressed in neurogenic areas and is also abundantly expressed in central and peripheral axons during axon outgrowth (e.g. dorsal root ganglia, trigeminal ganglion, motor neurons) (Escalante-Alcalde et al., 2009). Since LPA and S1P participate in neurogenesis as well as in the inhibition of neurite outgrowth, we hypothesize that LPP3 plays a fundamental role in regulating the activities of these lipids during early neural development.

Targeted inactivation of Lpp3 in the mouse renders embryonic lethality around day 9.5 of gestation (E9.5) (Escalante-Alcalde et al., 2003) due to severe defects in vascular system development. Although the embryos show evidences of alterations in nervous system (NS) development, the vascular phenotype makes difficult to attribute any alteration to an autonomous participation of LPP3 in the NS. On the other hand, the conditional inactivation of the Lpp3 gene in neural lineages using Nestin::Cre mice (López-Juárez et al., 2011) produces viable individuals with no apparent phenotype in the regions or neuronal cell types where LPP3 is expressed during early neural tube development, presumably due to the relatively late onset of Cre expression in that mouse line (E11) (Tronche et al., 1999; Chen et al., 2006). In this study we used in vitro differentiation of homozygous null ES cells (particularly into spinal derivatives including motor neurons) to address the participation of LPP3 in early neural development. Our results uncover a role for LPP3 in spinal neuron differentiation as well as in neurite outgrowth. In addition, we show that the lack of LPP3 in ES cells results in the appearance of a large proportion of smooth muscle actin positive cells under neural differentiation conditions, which was partially reverted by the treatment with a sphingosine kinase inhibitor. Hence, LPP3 appears to be an important regulator of neural differentiation and further study of this protein has the potential to offer new insights into the mechanisms involved in this process.

Results

The absence of LPP3 alters the formation of ES cell-derived spinal neural precursors and neurons

LPP3 is expressed in the spinal cord from its earliest stages of development. At E8.5-12.5 it is present in the ventral domains and is particularly enriched in motor neurons (MN). Between E12.5 and E15.5, its expression domain extends to the ventricular and subventricular regions (Escalante-Alcalde et al., 2009). To gain further insight into the role of LPP3 during early CNS development, we used in vitro differentiation of Lpp3^−/− ES cells as a model system. Lpp3^−/− and wildtype (wt) ES cells were subjected to a very robust and well established differentiation protocol (Wichterle et al., 2002), which enriches for spinal neurons by inducing caudalization of neural precursors differentiating within embryoid bodies (EB) through the addition of retinoic acid (RA) (Fig. 1).

Figure 1.

Differentiation protocol and treatments.

We first evaluated the effects of LPP3 deficiency during the initial steps of neural differentiation by analyzing the amount of neural precursors (Nestin⁺) formed in wt and Lpp3^−/− EB. Two days after RA treatment (day 4 of culture) EB were dissociated and the cells plated under adherent conditions and immunostained. As shown in figure 2A (top panels) and 2B, we noticed a 9-fold reduction in the proportion of neural precursors generated in Lpp3^−/− EB compared to wt controls. This observation was supported by the strong reduction in the expression of neuroectodernal markers, such as Sox1 and N-cadherin (Kamiya et al., 2010) in EB’s protein extracts (Fig. 2D and E). Absence of LPP3 protein in homozygous mutant ES cells was confirmed by western blot (Supplemental Fig. 1A).

Figure 2.

LPP3 deficiency results in reduced formation of neural progenitors and young neurons. A) Immunofluorescence for nestin and phospho-histone H3 (top panels) and β-III-tubulin (bottom panels) in wt and Lpp3^−/− cells. Percentage of nestin (B) and β-III-tubulin (C) positive cells. D) Expression of the neuroectodermal markers Sox1 and N-Cadherin in wt and Lpp3^−/− EB’s total protein extracts. EB were treated 2 days with RA. E) Normalized expression of markers in Lpp3^−/− EB relative to wt. F) Quantification of proliferating (pHH3⁺) neural precursors (Nestin⁺) in EB treated 2 days with RA. Only double-labeled cells (yellowish) were quantified. Scale bars=100 μm. * p= 0.02, *** p=0.0004

Next, we analyzed the number of differentiated neurons (β-III-tubulin⁺) in cells plated after 4 days of treatment with RA (Fig. 2A, bottom panels). As expected from the reduction in the proportion of neural precursors, Lpp3^−/− cells also generated a lower proportion of neurons (30%) than wt cells (80%) (Fig. 2C).

The lack of LPP3 also reduces the amount of differentiated neurons in teratomas

To test the neural differentiation potential of Lpp3^−/− ES cells in another system, we induced the formation of teratomas by injecting wt and Lpp3^−/− ES cells in athymic^nu/nu mice. The histopathological analysis (as reported by the Pathology & Histotechnology Lab, NCI-Frederick, MD, USA) revealed that teratomas formed by LPP3-deficient ES cells (n=4), were significantly smaller (3.4-fold) than their wt counterparts (n=5) and had an important reduction in the amount of differentiated neural tissue. To corroborate this observation we assessed the amount of βIII-tubulin⁺ tissue differentiated within the teratomas. As shown in figures 3A and 3B, despite the differentiation of multiple cell types, the amount of βIII-tubulin⁺ tissue in Lpp3^−/− teratoma’s sections was nearly 75% lower than in wt tumors.

Figure 3.

Lpp3^−/− teratomas show a reduction in the differentiation of neural tissue. A) Upper panels show histological sections from wt- and Lpp3^−/−-derived teratomas. Lower panels show the detection of neuronal tissue by β-III-tubulin immunostaining (purple) in sections counterstained with eosin (pink). B) Statistical analysis showing significant differences in the amount (% of marker positive area) of neuronal tissue present in wt vs. Lpp3−/− tumors. ep, epithelium; c, cartilage; m, muscle; nt, neuronal tissue *** p=0.0001. Scale bars= 200 μm, insets 20 μm.

LPP3-deficiency affects the in vitro formation of spinal neural precursors from ES cells by different mechanisms

Several mechanisms could account for the severe reduction in neural precursors observed: reduction in the proliferation rate, increased cell death and/or the precocious differentiation of neural precursors, or the preferential differentiation of ES cells to other cell lineages. Thus, we then set out to distinguish between these possibilities.

To evaluate neural precursor proliferation, EB treated for 2 days with RA were dissociated, plated and cells were identified by nestin and p-Histone-H3 immunostaining (pHH3). We found a 55% and 75% reduction in the amount of H³-thymidine incorporation and mitotic cells, respectively, in Lpp3^−/− vs. wt cells (Supplemental Fig. 1B and C). In addition, dividing double-positive nestin/pHH3 cells were around 18 times less abundant in Lpp3^−/− than in wt cells (Fig. 2A and F). These data indicated a clear perturbation of the proliferative capacity of LPP3-deficient cells and neural precursors. The stronger reduction in the number of proliferating neural precursors (Fig. 1F) compared to that found in the total population (Supplemental Fig. 1C) of Lpp3^−/− cells suggests that the former might be more sensitive to the lack of LPP3.

Throughout the differentiation process it was evident that both the size and integrity of the Lpp3^−/− EB were altered. To analyze this, we induced the formation of EB by the hanging drop method to tightly control their size. We observed that even from day 3 of culture (day 1 with RA) LPP3-mutant EB were smaller than wt controls, and that their superficial layers tended to dissociate. By day 6 of differentiation (day 4 with RA), mutant EB were 60% smaller than wt controls (Fig. 4A and B). Moreover, staining with the vital dye trypan blue revealed that many of the cells on the surface of mutant EB were dead (Fig. 4A). These alterations were not a consequence of the treatment with RA, since the same phenotype was observed when performing the differentiation in the absence of this factor (Supplementary Fig. 2A). These results suggested that LPP3 deficiency during ES cell differentiation compromises cell viability. TUNEL assays performed on sections of wt and mutant EB treated with RA for 2 or 4 days, suggested that apoptosis contributed importantly to the type of cell death triggered by the LPP3 deficiency (Figs. 4A and D). This observation was further supported by the increase in cleaved caspase-3 found in mutant EB (Supplemental Fig. 2B) and the significant increase in EB’s size (Fig. 4B), yield of neurons (Fig. 4C) and reduction in the proportion of TUNEL positive cells (Fig 4D) when Lpp3^−/− EB were treated with the general inhibitor of caspases Q-VAD. The increment in the number of differentiated neurons (15%) was not as efficient as the increment in the size of EB (250%) indicating that the population of rescued cells by Q-VAD was not restricted to the neuronal lineage. No effect was observed in EB’s size or neuron formation when differentiation was performed in the presence of the PI3K inhibitor LY294002 and/or of the ROCK inhibitor Y27632 (Supplemental Fig. 3) (Dottori et al., 2008).

Figure 4.

Lack of LPP3 produces a reduction in the amount of differentiated neurons from ES cells, partially through the increase in apoptotic cell death. A) From left to right, first column (EB D1) shows the appearance of EB after the first day of differentiation and the second column after 6 days of differentiation (EB D6). EB were stained with trypan blue. Third column shows differentiated young neurons (β-III-tubulin⁺, red) in each condition. Fourth column shows TUNEL positive cells in wt (top) and mutant EB without treatment (middle) or treated with Q-VAD (bottom). Quantification of EB’s highest diameter area (B), amount of differentiated neurons (C) and TUNEL⁺ tissue (D) after the sixth day of differentiation. Scale bars=100 μm except in trypan blue stained EB in which scale bar=500 μm.

We next assayed for premature appearance of differentiated neurons in stages where neural precursors should be majority (2 days with RA) and found no evidence of premature differentiation of this cell type (Figs. 5A top panels and B). Previous work in our laboratory suggested that Lpp3^−/− ES cells were predisposed to differentiate into α-smooth muscle actin positive cells (SMA⁺) when differentiated to vascular cell types. In addition, it has been reported that differentiation of mouse ES cells in the presence of S1P promotes the generation of cardiomyocytes, which transiently express SMA (Clement et al., 2007). Thus, we looked for SMA⁺ cells in our Lpp3^−/− ES cell differentiated cultures. Remarkably, LPP3-deficient EB treated with RA for 2 or 4 days contained a high proportion of SMA⁺ cells (Figs. 5A, C and E).

Figure 5.

Lack of LPP3 promotes the differentiation of SMA⁺ cells which is partially reverted by inhibition of sphingosine kinases. A) Immunofluorescence for of β-III-tubulin (red) and SMA (green) after 2 or 4 day of treatment with RA. Graphic representation of the amount of differentiated β-III-tubulin⁺ (B, D) and (C, E) SMA⁺ cells with or without treatment with DMS, and after 2 (B, C) and 4 (D, E) days of treatment with RA. F) Mass spectrometric analysis of DHS1P extracted from control and mutant EB’s conditioned media. Mean ± SD (n=5), t test *(p<0.05). Scale bars=100 μm.

To address if an unbalance in the concentration of lysophospholipids was responsible for the alterations observed, we performed experiments in the presence of inhibitors of sphingosine kinases or autotaxin. No modification of the phenotypes was observed when differentiation of mutant EB was performed in the presence of BrP-LPA, an antagonist of four LPA receptors and an inhibitor of the LPA synthesizing enzyme autotaxin (Prestwich et al., 2008).

On the other hand, when differentiation of Lpp3^−/− EB was performed in the presence of a low concentration of the inhibitor of sphingosine kinases N,N-dimethylsphingosine (1 μM, DMS), a significant increase in the yield of neurons with a concomitant reduction in the amount of SMA⁺ cells was observed (Fig. 5A-E). Interestingly, addition of DMS yielded a 3-fold increase in the amount of Lpp3^−/− differentiated neurons after 2 days of treatment with RA when compared to wt or inhibitor untreated Lpp3^−/− ES cells. This suggested that lowering S1P synthesis below a certain threshold could trigger a precocious differentiation of neurons in the Lpp^−/− background (Fig. 5A upper panels and B). Based on these results we used mass spectrometry to measure, by, the levels of secreted S1P and dihydro-S1P (DHS1P), both products of the Sphk activity, in conditioned medium of wt and Lpp^−/− EB treated 2 days with RA. We found a 2-fold increase in the amount of DHS1P, but not S1P, in conditioned medium of mutant EB (Fig. 5F), suggesting the participation of the extracellular accumulation of DHS1P in some of the observed phenotypes.

To test this hypothesis we differentiated wt ES cells in the presence of micromolar concentrations of DHS1P. Chronic treatment with exogenous DHS1P during the differentiation period produced a significant reduction in the size and viability of EB when compared to those treated with vehicle only (Supplemental Fig. 4). In agreement, the amount of neurons produced was smaller in treated EB. Also a mild but significant increase in the amount of SMA⁺ cells was observed in cultures treated with DHS1P (Supplemental Fig. 4). These results supported that accumulation of extracellular DHS1P could contribute to the reduction in EB’s size and viability and to the increase of SMA⁺ cells observed in LPP3-deficient cultures.

Altogether these data show that LPP3 deficiency alters the differentiation of ES cells to spinal neurons by a combination of mechanisms: reducing the proliferating capacity of neural precursors, increasing apoptotic cell death and promoting the differentiation of SMA expressing cells within the EB.

LPP3 is required for proper neurite outgrowth but not for MN lineage specification

The majority of Lpp3^−/− neurons differentiated in vitro failed to properly extend neurites (Fig. 2A). Since LPP3 is abundantly expressed in growing axons of MN during development (Escalante-Alcalde et al., 2009), we studied the effect of LPP3 deficiency in this particular cell type of spinal neuron. To this end, we differentiated Lpp3^−/− EB in the presence of RA and SHH, and MN differentiation was reported by the expression of a Hb9::EGFP reporter construct (Wichterle et al., 2002). We found that LPP3 deficient cells were able to differentiate to MN, as indicated by the expression of EGFP and the co-expression of Islet1/2 (Wichterle et al., 2002; Thaler et al., 2004), however they were unable to properly extend neurites (Fig. 6A). This result indicated that LPP3 is not required for MN lineage specification but suggested its participation in neurite extension.

Figure 6.

Lpp3^−/− ES cells differentiate to MN but have impaired neurite outgrowth. A) MN co-expressing Islet1/2 and EGFP were differentiated in the presence of RA and SHH. B) LPP3-deficient MN (green) and spinal young neurons (red) were treated with ROCK inhibitor (top panels) or PI3K inhibitor (bottom panels). Quantification of MN with extended neurites (C) and the length of neurites (D). Scale bars=100 μm.

Treatment with ROCK or PI3K inhibitors partially rescues neuritogenesis in LPP3 deficient neurons

LPP3 attenuates the effects mediated by LPA and S1P, whose neurite retractant activities rely on the activation of the Rho/ROCK pathway (Fukushima et al., 2002). Therefore, we explored whether inhibition of ROCK was able to rescue the defective neurite extension observed in LPP3 mutant neurons. Treatment of neurons obtained by dissociation of EB grown for 4 days with RA/SHH and cultured for further 24 hours in the presence of Y27632 (50 μM) partially rescued neurite outgrowth (Figs. 6B top panels and C). These data suggested that the failure in neurite outgrowth was partially due to increased ROCK activation in Lpp3^−/− neurons. Since it has been reported that PI3K can act in concert with ROCK to regulate axon extension and branching (Leemhuis et al., 2004), we performed the same kind of experiment but in the presence of the PI3K inhibitor LY297002. As illustrated in figures 6B (bottom panels) and D, inhibition of PI3K activity also promoted neurite outgrowth in Lpp3^−/− neurons. These results show that inhibition of ROCK and PI3K activities are able to partially overcome the neurite outgrowth deficit triggered by the lack of LPP3.

LPP3 deficient ES cells differentiated on wt neural tube embryo explants are unable to properly extend neurites

To determine if the neurite-outgrowth phenotype found in LPP3-deficient ES cell-derived neurons is due to enhanced extracellular lysophospholipid signaling or due to modifications in intracellular signaling acting in a cell-autonomous fashion, Lpp3^−/− ES cells were differentiated in the context of the wt neural tube. Dissociated cells from EB carrying the Hb9::EGFP reporter construct and cultured for 2 days were placed over the midline of “open-book” preparations of neural tube explants from E10.5 embryos and cultured for further 5 days. MN differentiation was evidenced by the expression of EGFP (Fig. 7). In accordance with our in vitro differentiation findings, LPP3 deficient cells differentiated to into MN in this ex vivo conditions, demonstrating that LPP3 is not required for MN lineage specification. However despite developing in a wt environment, LPP3-deficient neurons still showed neurite outgrowth deficits (Fig. 7). This result strongly suggested that lack of LPP3 produces additional alterations in intracellular signaling that contributes to the neurite outgrowth deficit.

Figure 7.

Lpp3^−/− ES cells differentiate to MN in ex vivo wt conditions but show abnormal neuritic outgrowth. Top panels show differentiated MN expressing EGFP in the ventral area of the wt neural tube. The dotted lines indicate the position of the midline. Scale bars=100 μm. Bottom panels show the morphology of in situ differentiated MN. Note that Lpp3^−/− MN have some cellular protrusions that are not as robust as wt MN (arrows). Scale bars=20 μm.

Discussion

Expression studies of lysophospholipid receptors during mouse and chick development suggest that lysophospholipid-mediated signaling plays important roles during early NS development (Meng and Lee, 2009; Ohuchi et al., 2008). However, single gene inactivation of the receptors has not been able to reveal the participation of any particular receptor in early NS development. On the other hand, gene inactivation of LPA and S1P metabolizing enzymes has profound effects on early NS development. Autotaxin (Atx) targeted inactivation suggested that it plays important roles in anterior brain morphogenesis and establishment of the MHB (Ohuchi et al., 2010; Koike et al., 2011). In addition Atx conditional inactivation in the epiblast produced defective neural tube closure, decreased neuroepithelium proliferation, increased cell death and reduction in the amount of differentiated young neurons (Tanaka et al., 2006; van Meeteren et al., 2006; Fotopoulou et al., 2010). Sphk1/Sphk2 double knockout show neural tube closure defects, increased cell death and reduced proliferation principally in the anterior neuroepithelium (Mizugishi et al., 2005). Embryos with LPP3 gene inactivation show anterior neural tube closure defects (Escalante-Alcalde et al 2007) and increased apoptotic cell death of the neuroepithelium (DEA, unpublished data). Moreover its expression pattern correlates with that of several lysophospholipid receptors and synthesizing enzymes such as in the notochord, midbrain-hindbrain area, floor plate, ventricular and subventricular zones and in neural crest derivatives (Escalante-Alcalde et al., 2003; Escalante-Alcalde et al., 2009), implying that it serves as a key regulator of these lysophospholipid mediated responses in the NS. Here we show that lack of LPP3 reduces the amount of neural precursors derived from differentiating ES cell cultures through a reduction in proliferation and increased apoptotic cell death. It is interesting that similar phenotypic characteristics to those produced by the deficiency of sphingosine kinases and autotaxin (lack of lipid synthesis) arise in a situation in which the concentration of substrate lipids is increased (Escalante-Alcalde et al., 2003; López-Juárez et al., 2011). This might suggest that receptor desensitization/down-regulation due to an increased lysophospholipid concentration could be at play as we had recently shown in the LPP3-deficient cerebellum (López-Juárez et al., 2011).

Alternatively, the unbalance in the intracellular metabolism of phosphorylated sphingolipids could be responsible for some of the observed phenotypes. According with previous studies de novo synthesis of DHS1P is mainly through Sphk1, but its formation and/or accumulation is not a major metabolic pathway in mammalian cells (Berdyshev et al., 2006). Thus, the significant increase of DHS1P, but not of S1P, that we detected in the conditioned medium of LPP3 deficient cultures is intriguing. If DHS1P synthesis is increased it could have two different outcomes (Berdyshev et al., 2006). When synthesized through the overexpression of Sphk1, it may interfere with exogenous S1P-mediated responses. On the other hand when it is synthesized by the addition of exogenous DHS, it would promote the increase of ceramide formation which in turn affects cell survival (Edsall et al., 1997; Hartfield et al., 1997; Shinpo et al., 1999). This raises the possibility that intracellularly located LPP3 and Sphk1 contribute to the regulation of intracellular DHS1P synthesis in this system, and that disruption of LPP3 activity produces alterations in the metabolism and biological activities mediated by this lipid. A detailed lipid and enzymatic profiling will be required to fully understand the role of LPP3 in the regulation of sphingolipid mediated responses and metabolism.

One remarkable finding of our work is the high proportion of SMA⁺ cells differentiated from Lpp3^−/− ES cells under neural-inducing conditions. This could be the result of the differentiation of this particular cell type from a population of mesodermal cells during the first steps of the differentiation protocol. As mentioned before, S1P-mediated signaling has been shown to promote cardiomyocyte differentiation in EB (Sachinidis et al., 2003) and cardiomyocytes transiently express SMA during early steps of differentiation (Clement et al., 2007). Alternatively an enhancement in the formation of cells expressing SMA of neural crest origin (pericytes, cardiac neural crest, myofibroblasts) might occur. TGF®-family members have a potent differentiation effect on neural crest and neural stem cells towards a smooth muscle fate (Shah et al., 1996; Sailer et al., 2005), and it is well established that there is crosstalk between S1P and the TGF® signaling pathway (Xin et al., 2004). It will be interesting to determine if SMA⁺ cells differentiated in Lpp3^−/− cultures are indeed neural crest-like and if their appearance is regulated through the S1P-mediated transactivation of SMAD effectors. Whichever is the mechanism, it is clear that the appearance of SMA⁺ cells in Lpp3^−/− cultures is triggered by the increase of phosphorylated sphingolipids, since differentiation in the presence of 1 μM DMS (the inhibitor of sphingosine kinases) reduced the amount SMA-expressing cells. In addition wt ES cells differentiated under neural-inducing conditions in the presence of high concentrations of DHS1P increases the amount of SMA expressing cells, although the increase was not comparable to that found in LPP3-deficient cells presumably due to the lipid dephosphorylating activity present in wt cells.

Finally, we were able to reveal the participation of LPP3 in regulating neurite and axon outgrowth of in vitro differentiated spinal neurons through the regulation of ROCK and PI3K activities. Neurite and axon outgrowth are regulated by a strict balance between the activities of small GTPaes of the Rho family (Govek et al., 2005). Rho/ROCK activation promotes inhibition of neurite initiation and retraction while activation of Rac1 promotes neurite outgrowth, although in several instances it has the opposite effect. Neurite retraction in several neuronal types and cell lines is mediated by the activation Rho/ROCK through the activation of LPA and S1P receptors coupled to G_12/13 proteins (Postma et al., 1996; Fukushima and Morita, 2006). On the other hand, Rac1 is one of the main effectors of PI3K, which can be activated by a wide variety of LP receptors (Sanchez and Hla, 2004; Choi et al.). However, the inability of differentiated LPP3-deficient spinal neurons to properly extend neurites in a wt environment, strongly suggests that this particular phenotype is due to the modification of intracellular LPP3-mediated responses or phosphatase independent activities (Brindley and Pilquil, 2009).

Since LPP3 is abundantly expressed in growing axons during early mouse development it is possible that it participates in axon outgrowth and guidance during early NS development in vivo. Further research conditionally inactivating LPP3 gene in the earliest stages of neural development will be helpful to determine the roles of this particular enzyme in early neuroepithelium development and axon outgrowth and guidance.

Experimental Procedures

ES culture and differentiation protocol

W9.5 and Lpp3−/− mouse ES cells (Escalante-Alcalde et al., 2003) were co-electroporated with linearized Hb9::EGFP construct, to report for motor neuron (MN) differentiation (Wichterle et al., 2002), and a PGK-hygromycin plasmid. Hygromycin B resistant clones were screened for the expression of EGFP under MN differentiation conditions. ES cells were expanded on Mitomycin C-treated primary mouse embryonic fibroblasts, with DMEM-high glucose supplemented with 0.1 mM nonessential amino acids, 2 mM GlutaMax, 1 mM sodium pyruvate and 50 UI/mL penicillin, 50 μg/mL streptomycin (all reagents from Invitrogen), 0.1 mM 2-mercaptoethanol (Sigma-Aldrich) and 15% ES cell tested fetal bovine serum (FBS; Wisent, Canada).

For differentiation, ES cells were grown to 80% confluency whereupon they were trypsinized and cultured in suspension in bacteriological plates to induce embryoid bodies (EB) formation. EB were cultured in DNFK differentiation medium, consisting of DMEM:F12 and Neurobasal medium (1:1) supplemented with 0.1 mM 2-mercaptoethanol, 2 mM GlutaMax, 50 UI/mL penicillin, 50 μg/mL streptomycin and 10% knockout serum replacement (Invitrogen). After 2 days of culture, DNFK medium was supplemented with 2 μM retinoic acid (Sigma) and EB were cultured for further 4 days to induce caudal spinal neural differentiation. For MN differentiation assays, DNFK medium was supplemented with RA and 100 ng/mL of recombinant human SHH (R&D Systems). For analyses, EB were dissociated using 0.25% trypsin and cells (500,000 cells per well) were cultured for 24 hrs into 4-well plates, pre-treated with 1 μg/ml laminin, 5 μg/ml fibronectin and 15 μg/ml poly-L-ornithine. The number of neural progenitors and proliferation were measured at day 4 (after 2 days with RA treatment). The amount of young neurons with neurite outgrowth was analyzed at day 6 (after 4 days with RA treatment). In some experiments, 1 μM N,N-Dimethylsphingosine (DMS; Cayman Chemical Company), 2 μM Q-VD-OPH (QVAD; MP Biomedicals), 0.1–10 μM D-erytrho-dihydro sphingosine-1-phosphate (DHS1P; Sigma) were added to the media from the beginning of EB formation. In other cases, cells were treated with 50 μM of Y27632 (Calbiochem) or 10 μM of LY294002 (Calbiochem) for 24 hrs after EB dissociation.

For hanging drop assays, dissociated ES cell cultures were seeded onto 0.1% gelatin-treated plates for 1 hr to eliminate feeder cells. ES cells, which remained in suspension, were then diluted at 20,000 cells/ml in DNFK medium and 20 μL drops were placed on the up-turned inner surface of a 100 mm tissue culture dish lid. The lid was then carefully inverted and placed on top of the dish containing 5ml of DNFK medium to prevent the drops from drying. Hanging drops allowed the formation of EB of uniform size, which were then expanded in suspension culture as described above.

Immunofluorescence

Cells were washed 3 times with PBS, fixed with 4% paraformaldehyde for 5 min, rinsed 3 times with PBS and incubated in blocking solution (0.1% Triton X-100 and 1% FBS in PBS) for 20 min at room temperature. Cells were then incubated with primary antibodies diluted in blocking solution overnight at 4°C. The following day, cells were washed with PBS and then incubated with Alexa fluor-conjugated secondary antibodies (Molecular Probes, Invitrogen) diluted in blocking solution for 1 h at room temperature. After washes, cells were stained with a 1 μg/ml of DAPI for 10 min and then washed before visualization. The primary antibodies used were: anti-nestin (1:100; mouse; Chemicon), anti-phospho-histone H3 (1:100; rabbit; Cell Signaling), anti-β-III-tubulin (1:500; rabbit; Sigma), Islet 1/2 (supernatants 39.4D5 y 40.2D6 1:1; mouse; Developmental Studies Hybridoma Bank), anti-α-smooth muscle actin (SMA; 1:200; mouse; Oncogene).

Differentiation in cultured explants

Explant cultures were performed as previously described (Baizabal and Covarrubias, 2009) with some modifications. EB were generated as described above. After 2 days of culture in DNFK medium, they were dissociated by a 10 min treatment with 0.25% trypsin at 37°C. Trypsin was inactivated by addition of 10% SFB/DMEM and cells were gently resuspended to get a homogeneous cell suspension, spun down at 3000 rpm for 4 min and resuspended in 20 μl of DNFK medium. This cell suspension was applied over E10.5 “open book” neural tube explants (at the level of the forelimbs), which were cultured on floating polycarbonate membranes (Millipore). Cells were allowed to adhere to the explant for 12 hrs, removed from the filter and then embedded in a collagen matrix with the ventricular surface facing up and cultured in explant media. The collagen mixture contained: rat collagen (100 μl), 1.5 M NaCl (10 μl), 7.5 % NaHCO3 (10 μl) and explant media (300 μl). Explant media was composed by Optimem (72% v/v), DMEM: F12 (25% v/v), Glucose 2 M (2% v/v), GlutaMax 2 mM and UI/mL penicillin, 50 μg/mL streptomycin (all reagents from Invitrogen). Collagen-embedded explants were incubated at 37°C for 1 hr to allow collagen polymerization; explant media was then added to fully cover the tissues. Explants were cultured for further 5 days.

Data Analysis

All experiments were independently repeated at least three times. Statistics were determined via Student’s t-test (when comparing two groups) or ANOVA (when comparing more than 2 groups) using GraphPad Prism 5. P values <0.05 as compared with control cells were considered to be statistically significant. β-III-tubulin⁺ tissue and EB’s highest diameter area was calculated using Image J sofware. Cell percentages were calculated counting individual cells in at least 5 fields per experiment.

Western blot analysis

EB were homogenized in lysis buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1% NP-40) containing protease inhibitors (Roche) and phosphatases inhibitors (1 mM NaVO4 and 10 mM NaF). Thirty to eighty μg of total protein were separated on 10–12% SDS-PAGE and transferred to PVDF membranes (Hybond-P, Amersham Pharmacia GE Healthcare). Membranes were incubated with the appropriate primary and secondary antibody dilution. Immunoblots were developed by ECL. Densitometric analysis was done using Image J software (NIH), data normalized against GAPDH and expressed as relative protein levels versus controls.

Quantification of Glycero- and Sphingo-Phospholipids by High-Pressure Liquid Chromatography and Tandem Mass Spectrometry (LC-MS/MS)

Lipids from EB’s conditioned medium at day 2 with RA treatment were extracted in the presence of 50 pmol of C17 LPA (Avanti Polar Lipids) as a recovery standard. Samples were evaporated to dryness and redissolved in 4:1 MeHO; CHCl₃ for determination of phospholipids phosphorous and analysis by HPLC electrospray ionization tandem mass spectrometry using selective reaction monitoring mode assays and methods that have been described elsewhere to measure PA, LPA and S1P species (Su et al., 2009; Albers et al., 2010; Mathews et al., 2010). Data were corrected for recovery of C17 LPA internal standard and lipid species were quantitated by reference to calibration curves generated using synthetic standards (Avanti Polar Lipids) that were independently quantitated by lipid phosphorous determination. The data shown are means of 5 independent measurements made with independent samples.