Sphingosine kinase 1/S1P receptor signaling axis is essential for cellular uptake of Neisseria meningitidis in brain endothelial cells

Invasion of brain endothelial cells (BECs) is central to the pathogenicity of Neisseria meningitidis infection. Here, we established a key role for the bioactive sphingolipid sphingosine-1-phosphate (S1P) and S1P receptor (S1PR) 2 in the uptake process. Quantitative sphingolipidome analyses of BECs infected with N. meningitidis revealed elevated S1P levels, which could be attributed to enhanced expression of the enzyme sphingosine kinase 1 and its activity. Increased activity was dependent on the interaction of meningococcal type IV pilus with the endothelial receptor CD147. Concurrently, infection led to increased expression of the S1PR2. Blocking S1PR2 signaling impaired epidermal growth factor receptor (EGFR) phosphorylation, which has been shown to be involved in cytoskeletal remodeling and bacterial endocytosis. Strikingly, targeting S1PR1 or S1PR3 also interfered with bacterial uptake. Collectively, our data support a critical role of the SphK/S1P/S1PR axis in the invasion of N. meningitidis into BECs, defining a potential target for adjuvant therapy.


Introduction
Sphingolipids are ubiquitous plasma membrane constituents, but can also act as bioactive molecules in eukaryotic cells [1].Both functions are essential for cellular homeostasis and dysregulation can contribute to various conditions, such as Alzheimer's disease, Parkinson's disease, Multiple Sclerosis or infectious diseases [2].Many pathogens, including bacteria, viruses and fungi, manipulate cellular membrane sphingolipids and metabolites to enhance their virulence [3,4].Some extracellular bacterial pathogens can use ceramide-enriched membrane platforms (CRPs) as an entry portal into various host cells [5][6][7].Other microbes, such as obligate intracellular Chlamydia trachomatis, are able to capture host sphingolipids and incorporate them into their membranes [8].The β-toxin-producing bacterium Staphylococcus aureus can actively target host sphingolipids by hydrolyzing sphingomyelin [9].Furthermore, some bacteria e.g.members of the Bacteroidetes phylum are capable of synthesizing sphingolipid species in order to survive in hostile environments and improve replicative fitness [10].
Sphingosine-1 phosphate (S1P) is a sphingolipid metabolite that acts as a bioactive signaling molecule.S1P is formed by phosphorylation of sphingosine (Sph) by two isoforms of sphingosine kinases, SphK1 and SphK2 [11].SphK1 activation can be regulated at the transcriptional, translational or post-translational level by phosphorylation and protein-protein interactions which can control both the activity and localization [12].Phosphorylation of SphK1 at Ser-225 for example leads to the translocation of the kinase to caveolin-coated lipid rafts, where the enzyme encounters Sph and is thus stimulated to form S1P [13,14].While SphK1 is found in the cytoplasm [15], SphK2 is predominantly found in the nucleus, where S1P production can directly influence cellular transcription by manipulating telomerase reverse transcriptase or histone deacetylation [16,17].
S1P is exported out of endothelial cells predominantly by the spinster homology protein 2 (Spns2) [18].Extracellularly, S1P modulates a variety of cellular processes via binding to a family of five specific G protein-coupled receptors (S1PR1-5).Of these, S1PR1-3 are ubiquitously expressed, while S1PR4/5 expression is mostly restricted to hematopoietic cell types [19].S1P is involved in regulating cardiovascular functions, neural development and immune cell transport [20].S1P also modulates angiogenesis and stability of the vasculature [21].This also applies for the brain microvasculature, where the balance between S1PR1/3-and S1PR2-signaling ensures the stability of the blood-brain-barrier (BBB) [22][23][24][25].Recent findings attribute this effect not only to paracellular but also transcellular integrity [26].
Neisseria meningitidis (N.meningitidis) are commensal inhabitants of the human nasopharynx [27] and it is likely that most people are colonized at least once throughout their life time [28].In rare cases N. meningitidis can cross the nasopharyngeal epithelial barrier to reach the bloodstream and cause severe septicemia and/or meningitis.The binding of bacteria to endothelial cells of the brain (BECs) and their invasion is a prerequisite for successful invasion of the cerebrospinal fluid, thus causing meningitis.N. meningitidis possesses a variety of surface virulence factors that contribute to BEC adhesion, including type IV pili (Tfp), the outer membrane protein OpcA [29], lipopolysaccharides and a number of so-called minor adhesion proteins [30].Among them, Tfp play the most important role, enabling binding to endothelial receptor CD147/Basigin [31,32] and β-adrenergic receptor [33] thereby mediating a signaling cascade that leads to the formation of membrane protrusions around the adherent bacteria

PLOS PATHOGENS
SphK1/S1PR axis is essential for uptake of N. meningitidis in BECs and increases their ability to resist the mechanical forces generated by blood flow [34].Tyrosine phosphorylation of host proteins occurs after N. meningitidis binding to BECs, leading to cytoskeletal remodeling with subsequent engulfment of the pathogen.Specifically, the two non-receptor tyrosine kinases, c-Src and focal adhesion kinase, as well as the epidermal growth factor receptor (EGFR) are activated to mediate bacterial uptake [35][36][37].
Of note, while N. meningitidis is not capable of synthesizing sphingolipids itself, it is able to modulate the sphingolipid content on BECs, depending on the expression of Tfp and OpcA [6,7].Mechanistically, N. meningitidis transiently activates acid sphingomyelinase (ASMase) and thereby induce increased ceramide (Cer) levels on the plasma membrane of BECs, leading to the formation of large CRPs that serve to reorganize and bundle receptor molecules and allow bacterial entry into the cells [6].Cer levels increase only transiently under infection, consistent with further metabolization.However, nothing is yet known about the involvement of other sphingolipid metabolic enzymes in N. meningitidis infection.
Here, we profiled the sphingolipid metabolome using LC-MS/MS measurements to provide a time-resolved image of sphingolipid metabolites in BECs in response to infection with N. meningitidis.We observed a significant increase of S1P release that was due to enhanced abundance of SphK1 transcript as well as enzyme activity in response to N. meningitidis infection.Increased activity was found to be dependent on Tfp interaction with its cognate receptor CD147.We showed that N. meningitidis infection increased S1PR2 expression, leading to the activation of the S1P-S1PR2-EGFR axis and ultimately to bacterial uptake.Activation of S1PR1+3 however reduced bacterial uptake.This study demonstrates tight regulation of the S1PR2 and S1PR1+3 signaling on BECs disrupted during N. meningitidis infection by a novel criss-cross transactivation mechanism involving the three individual receptor systems CD147, S1PR2 and EGFR and illustrates the possible therapeutic potential of targeting this signaling cascade in meningococcal meningitis.

N. meningitidis modulates the sphingolipid metabolome in BECs
To examine whether in addition to ASMase, infection of BECs with N. meningitidis also regulated the activity of other sphingolipid metabolic enzymes, we first conducted a sphingolipid metabolome time-resolved survey using LC-ESI-MS/MS measurements.Targeted quantitative measurement of sphingolipids was performed at the indicated time points in brain microvascular endothelial hCMEC/D3 cells either left uninfected (control) or infected with N. meningitidis strain MC58 (Fig 1A ) or in their supernatants (Fig 1B).
Data analysis showed a moderate increase in cellular Cer in hCMEC/D3 and the differences did not reach statistical significance (Fig 1A).When comparing the extracellular levels of Cer, enhanced levels in Cer were observed in the supernatants of infected hCMEC/D3 (Fig 1B ), suggesting that excessively produced Cer was not fully retained in the plasma membrane.Interestingly, increase of major Cer species 16: 0, 24:0 and 24:1 was mostly observed extracellularly, whereas the low abundant species Cer species 18:0 and 20:0 Cer were prominently increased in both compartments (

N. meningitidis transiently activates SphK1 expression and activity
To determine whether S1P metabolic enzymes (Fig 2A) contribute to the increase of S1P levels we measured the levels of their mRNA transcripts in hCMEC/D3s infected with N. meningitidis MC58 over 8h infection time course using quantitative real time PCR (qRT-PCR).SphK1 mRNA levels were significantly upregulated in infected hCMEC/D3 at 4h p.i. compared to uninfected control cells (Fig 2B ), while SphK2 transcript levels were not changed and even decreased at 6h and 8h p.i. (Fig 2C).No significant changes in mRNA levels of degradative enzymes S1P phosphatase 1 and 2 (SGPP1, SGPP2) and S1P lyase (SGPL1) were detected (Fig 2D -2F).Likewise, N. meningitidis infection did not affect the expression of Spns2, the transporter mainly responsible for the secretion of S1P from endothelial cells (Fig 2G) [18].
We next measured SphK activity in BECs.Infection with N. meningitidis significantly increased SphK activity by 2h and remained elevated thereafter (Fig 2H), thus demonstrating that N. meningitidis infection alters abundance of SphK1 transcript as well as enzymatic activity.Phosphorylation of SphK1 at Ser-225 induces SphK1 activation through specific interaction with phosphatidylserine resulting in recruitment to the plasma membrane [14].Infection of hCMEC/D3 with N. meningitidis MC58 slightly increased phosphorylation levels of SphK1 at Ser-225 in whole cell lysates, albeit not significantly (Fig 2I).We therefore aimed to detect p(Ser-225)-SphK1 translocation by purifying plasma membranes from hCMEC/D3s, which were infected with N. meningitidis MC58 for 4h or left uninfected.Western Blot analysis showed, that SphK1 protein levels were increased and p(Ser-225)-SphK1 was only detectable in the plasma membrane fraction after infection, indicating increased phospho-activation and translocation of SphK1 upon meningococcal infection (S1B Fig) .We also examined a possible activation of SphK2 through the ERK1 phosphorylation site Thr-578 [38].However, p(Thr-578)-SphK2 levels were not changed in N. meningitidis-infected hCMEC/D3s (Fig 2I ).

N. meningitidis Tfp binding to CD147 contributes to SphK activity
Type IV pili (Tfp) of N. meningitidis trigger translocation and activation of ASMase in BECs [7].To examine whether Tfp also contribute to SphK activation, hCMEC/D3s were infected with wildtype strain N. meningitidis MC58 (Fig 3A ), mutant strain MC58 pilE, lacking the major constituent of Tfp PilE, or mutant strain MC58 pilT, lacking the pilus retraction machinery [39], leading to a hyper-piliated state of the ΔpilT mutant [40].While MC58 pilE mutant had no effect, the hyper-piliated MC58 pilT mutant elicited a significant increase of SphK activity (Fig 3A).To further corroborate the role of Tfp, we next purified pili from N. meningitidis strain 8013/clone12 [41], a meningococcal isolate that is highly piliated but lacks Opa and OpcA expression (Fig 3B).hCMEC/D3s were treated with pilus-enriched fractions (PeF) and PeF from an isogenic pilus-deficient mutant served as control.PeF from wildtype strain 8013/12 induced a significant increase of SphK activity 10 min after treatment, whereas PeF from the pilus-deficient mutant 8013/12 pilE did not (Fig 3B).
Attachment of Tfp to the endothelium is mediated by binding to the proximal extracellular immunoglobulin (IgG) domain of cellular surface receptor CD147 [32].To investigate, whether binding to CD147 leads to SphK activation, we tested two antibodies specific for CD147, MEM-M6/1 and MEM-M6/6, which bind to the extracellular N-terminal and the Cterminal Ig domains of CD147, respectively (Fig 3C) [31].Blockade of the membrane-distant domain with anti-CD147 Ab (MEM-M6/1) did not interfere with SphK activation, whereas competitive blockage of the binding domain with anti-CD147 Ab (MEM-M6/6) prevented PeF-induced SphK activation (Fig 3D).In addition, pre-treatment with ASMase inhibitor Amitriptyline [6] similarly prevented PeF-induced SphK activation (Fig 3E).Taken together, these results indicate, that N. meningitidis activates SphK in hCMEC/D3s potentially downstream of ASMase activation and dependent on the important virulence factor Tfp, and activation implicates CD147 as an important receptor.

N. meningitidis induces S1PR2-mediated EGFR activation to enable its uptake by hCMEC/D3s
The sphingolipid metabolome survey revealed a continuous increase of extracellular S1P released from hCMEC/D3s upon infection with N. meningitidis (Fig 1A and 1B).To establish the functional role of increased extracellular S1P and its receptors, we first determined meningitidis MC58 for 8h infection time course and both total protein and/or phosphorylated form of SphK1 (pSer-225) and SphK2 (pThr-578) were assessed by Western blot.Representative immunoblot is shown.Graphs represent the mean ± SD of three independent experiments and reported as phosphorylated form relative to total target protein normalized to control cells.One-Way ANOVA showed no significance.Related to S1 Fig. https://doi.org/10.1371/journal.ppat.1011842.g002

PLOS PATHOGENS
SphK1/S1PR axis is essential for uptake of N. meningitidis in BECs expression of S1PR1-5 in hCMEC/D3.S1PR1-3 were highly expressed in uninfected hCMEC/D3 cells, with negligible expression levels of S1PR4 and S1PR5 (Fig 4A).To evaluate the effects of N. meningitidis infection on the expression of S1PRs, hCMEC/D3s were infected for a period of 8h.Neither the mRNA levels of S1PR1 nor S1PR3 were affected by infection with N. meningitidis MC58 (Fig 4B and 4D).By contrast, mRNA levels of S1PR2 in hCMEC/D3 increased significantly (Fig 4C ).
Since previous studies have shown that uptake of N. meningitidis by BECs is mediated by activation of EGFR through phosphorylation at Tyr-845 [35] and a cross-talk between EGFR and S1P signaling has been reported [42], we sought to investigate a possible link between S1P, S1PR2 and the activation of EGFR in BECs in response to N. meningitidis infection.S1P treatment increased protein levels and Tyr-845 phosphorylation of EGFR (Fig 4E

PLOS PATHOGENS
SphK1/S1PR axis is essential for uptake of N. meningitidis in BECs

PLOS PATHOGENS
SphK1/S1PR axis is essential for uptake of N. meningitidis in BECs infection and even declined at 8h p.i., suggesting a partial turnover of EGFR at later time points of infection following EGF-mediated EGFR activation [43].
To further test, whether S1PR2 is involved in EGFR phosphorylation, hCMEC/D3 were treated with JTE-013, a S1P receptor antagonist, highly selective for S1PR2 [44].JTE-013 exhibited a dose-dependent inhibition of phosphorylation of EGFR measured in hCMEC/D3 cell lysates infected with N. meningitidis for 6h in a dose dependent manner.While treatment with 1 μM JTE-013 greatly prevented infection-induced EGFR phosphorylation, 10 μM JTE-013 reduced EGFR phosphorylation even below activation levels of uninfected control cells (Fig 4G).
EGFR activation has been linked to c-Src activation [35] which regulates cytoskeletal rearrangement known to be essential for N. meningitidis uptake in BECs [37].We therefore determined the role of S1PR2 in N. meningitidis invasion of hCEMC/D3 using JTE-013 at a concentration of 1 μM (S4 Fig) .Only 1 μM was tested, as it was recently reported that treatment with 10 μM JTE-013 has significant off-target effects on SphKs [45]

Targeting S1PR1 and S1PR3 interferes with N. meningitidis invasion into BECs
Although expression of S1PR1 and S1PR3 was not altered during the 8h infection time course (Fig 4B and 4D), both receptors may be involved in the interaction of N. meningitidis with BECs after the release of S1P.Targeted siRNA knockdown decreased mRNA expression of S1PR1 by 77% (S2A Fig

PLOS PATHOGENS
Transient knockdown of S1PR3 surprisingly led to an increase in intracellular bacteria, whereas knockdown of S1PR1 had no effect on N. meningitidis invasion into BECs (Fig 5B).
To further investigate the contribution of S1PR1/3 to bacterial uptake, we then examined the effects of various S1PR1/3 receptor modulators (summarized in Fig 5A).All antagonist/ agonists were used at concentrations, that were not cytotoxic and did not affect bacterial growth (S5 Fig) .hCMEC/D3s were treated with the solvent alone or with 10 μM of S1PR1-specific antagonist W146 [47] for 30 min prior to infection with N. meningitidis MC58.In contrast to the results from the S1PR1 knockdown experiments, pharmacological targeting of S1PR1 using W146 resulted in an increased number of intracellular bacteria (Fig 5C ), suggesting that signaling via S1PR1 and S1PR3 tend to prevent bacterial uptake.To further corroborate this finding, we assessed the effect of FTY-720, a non-selective agonist for 4 of the 5 known S1P receptors, S1PR1,3-5.Treatment of hCMEC/D3s with FTY-720 resulted in significant reduction of bacterial uptake (Fig 5D ).
FTY-720 is phosphorylated by sphingosine kinase 2 (SphK2) to function as a S1P receptor ligand [48], but was also found to inhibit SphK1 [49].To confirm that the reduction in bacterial invasion caused by FTY-720 was due to interaction with the receptor rather than interference with intracellular signaling from SphKs, we next examined the effects of the phosphorylated form of FTY-720, FTY-720-P.As observed for the treatment of hCMEC/D3s with FTY-720, pre-treatment with FTY-720-P resulted in significant reduction of bacterial uptake (Fig 5E).
To assess the individual role of S1PR1 CYM5442, a highly selective S1PR1 agonist, was tested [50].Treatment of hCMEC/D3s with CYM5442 significantly reduced the number of intracellular N. meningitidis throughout the course of infection (Fig 5F).We also tested the S1PR3-specific agonist CYM5541 to identify the contribution of S1PR3 [51].S1PR3-specific agonist CYM5541 also reduced the number of intracellular N. meningitidis (Fig 5G ) although this effect was not as pronounced as with the S1PR1 agonist.
To determine whether S1PR1/3-coupled G proteins transactivate EGFR, hCMEC/D3s were pre-treated with the agonists or antagonists used above, then infected with N. meningitidis for 8h and cell lysates were analyzed for p(Tyr845)-EGFR phosphorylation.Even though S1PR1+3 agonists showed a tendency to reduce EGFR phosphorylation, analyses of Western blots did not show significant effects upon S1PR1/3 treatment (Fig 5H).Taken together our results suggest that activation of S1PR1 or S1PR3 signaling prevents N. meningitidis invasion into BECs.However, this effect cannot be explained by the prevention of EGFR phosphorylation, hence indicating that S1PR1+3 signaling reduces N. meningitidis uptake by a different mechanism.

SphKs control N. meningitidis invasion into BECs
Having shown the ability of N. meningitidis to activate SphK in brain endothelial cells and that the resulting increase in S1P release triggers bacterial uptake through the activation of S1PR2 signaling, we wanted to confirm the specific role of SphKs in controlling N. meningitidis infection.SPHK1 expression was reduced by more than 57% after treatment with siRNA targeting SPHK1, however it also caused significant upregulation of SPHK2 (S3A Fig) .Similar, siRNA targeting SphK2 resulted in reduced SPHK2 expression by more than 70% and upregulation of highly expressed SPHK1 (S3B Fig) .We therefore set out to simultaneously knock down both isoforms.Double knockdown of both SphKs resulted in reduced levels of SPHK1 mRNA by more than 70% and SPHK2 mRNA by more than 80%, respectively (S3C Fig) .Remarkably, the double knockdown of SPHK1 and SPHK2 expression resulted in a significant inhibition of bacterial uptake at 8h p.i. (Fig 6A).To further determine which isoform is involved in interference with bacterial uptake we used several pharmacological inhibitors that specifically either

PLOS PATHOGENS
SphK1/S1PR axis is essential for uptake of N. meningitidis in BECs inhibit SphK1 or SphK2.All inhibitor concentrations tested had no cytotoxic effects on hCMEC/D3 or had any effect on bacterial growth (S6 Fig) .Treatment of hCMEC/D3s with PF543, a potent and specific inhibitor of SphK1 [52], greatly reduced the number of intracellular N. meningitidis over an 8h infection time course (Fig 6B).We next explored the effect of hCMEC/D3 treatment using two highly selective and potent SphK2 inhibitors, SLM6031434 [53] or K145 [54] (Fig 6C and 6D).While treatment with the inhibitors did not affect bacterial adherence (S6 Fig), the numbers of intracellular N. meningitidis significantly lowered, although this effect was not as pronounced as with SphK1 inhibition.

Disscussion
In this study we report a significant increase of S1P in BECs infected with N. meningitidis that correlated with increased SphK1 expression and activity.Increased SphK activity was found to be dependent on Tfp interaction with its cognate receptor CD147.S1P in turn was exported out of cells and activated S1PR2, whose expression was increased during infection.S1PR2 signaling activated EGFR and thus regulated bacterial uptake.In agreement, inhibiting SphK or S1PR2 activation markedly reduced N. meningitidis invasion into BECs.Collectively, our findings disclose a crucial role of S1P and S1PR2 in N. meningitidis invasion and establish a novel signaling mechanism, resembling criss-cross transactivation, induced by the pathogen, leading to SphK1 activation, mediating signals between three individual ligand-receptor systems (Tfp/ CD147, S1P/S1PR2 and EGFR).
Previous work from our group reports that the pathogen also transiently activates the enzyme ASMase, that hydrolyses sphingomyelin to ceramide and phosphocholine, thus increasing levels of ceramides on the plasma membrane of BECs [6,7].Surface ceramides form large ceramide-enriched membrane platforms (CRPs) that serve to reorganize and cluster receptor molecules and N. meningitidis uses such CRPs as entry portal [6,7,55].It is noteworthy that the activation of the enzyme ASMase can be attributed to the expression of Tfp [7], as we also showed here in this study for SphK activity.Moreover, SphK activation could also be prevented by ASMase inhibition.Two endothelial cell receptors are important for Tfp, CD147, a member of the immunoglobulin (Ig) superfamily, and the β2-adrenergic receptor [31,32,56].N. meningitidis first interacts with the membrane proximal IgG domain of CD147.
Here we delineated that the interaction between Tfp and the membrane proximal IgG domain of CD147 is necessary for SphK activation.
The use of LC-MS/MS measurements taken at different time points allowed us to establish a time-resolved profile of the sphingolipid pathway and thereby allowed us to quantify the formation of ceramides during N. meningitidis infection of BECs.Moreover, we observed a significant time-dependent increase of cellular sphingosine and S1P, peaking 2h and 4h p.i., One-way ANOVA followed by Dunnett's post-hoc test, ** p < 0.01.Absence of cytotoxicity and knockdown efficiency were verified as shown in S2 Fig. (C-G) hCMEC/D3 were pre-treated with S1PR antagonist W146 (C), S1PR1+3-5 agonist FTY-720 (D), S1PR1+3-5 agonist FTY-720-P (E), S1PR1 agonist CYM5442 (F) or S1PR3 agonist CYM5541 (G) or solvent control (DMSO) for 30 min and were subsequently infected with N. meningitidis MC58 for 8h.Numbers of viable intracellular bacteria were determined by gentamicin protection assay and normalized to control cells (DMSO treated cells).Data are means ± SD; n � 3, N = 2. Multiple t-test and p-value adjustment with Holm-Sidak correction.* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.(H) Assessment of pharmacological modulation of S1PR1 and S1PR3 on EGFR protein levels and phosphorylation at Tyr-845.hCMEC/D3s were pretreated with FTY720, W146, CYM5442 or CYM5541 or solvent control (DMSO) and infected with N. meningitidis MC58 or left uninfected for 8h.Cell lysates were analyzed by Western blot and subsequent densitometric analysis.Data represent means ± SD, n = 4. One-way ANOVA followed by Dunnett's post-hoc test showed no significance.Absence of cytotoxicity and bacteriostatic effects were verified as shown in S5 Fig. respectively.The transient elevation in sphingosine and S1P induced by N. meningitidis suggests that ceramide is subsequently degraded by ceramidase to produce sphingosine, the substrate for SphK.Sphingosine may also play a role in bacterial infection.Of note, dihydroceramides were also elevated in the cellular compartment as well as in the supernatants.Moreover, further studies are needed to determine how S1P, ceramides or dihydroceramides are released from hCMEC/D3s and whether hCMEC/D3s release extracellular vesicles upon N. meningitidis infection.
Importantly, N. meningitidis induced increased expression of S1PR2 in BECs, suggesting a shift in the S1PR expression pattern on hCMEC/D3 cells after meningococcal infection towards S1PR2-mediated signaling that is crucial for barrier disruption.Blocking of S1PR2 signaling resulted in impaired EGFR phosphorylation at Tyr-845 and abrogated bacterial uptake.EGFR can be activated downstream of G protein-coupled receptors (GPCRs) by two different mechanisms.The mechanisms involved in GPCR-induced EGFR signaling can occur through either the intracellular or extracellular pathways [57].Intracellularly, G proteins transactivate EGFR via Src-mediated phosphorylation at Tyr-845; extracellularly, released EGFR-like ligands stimulate EGFR directly.We previously reported that N. meningitidis induce Src-mediated phosphorylation of EGFR at Tyr-845 and this step is involved in the uptake process [35].Furthermore, we found that the EGFR is activated by the release of the ligand HB-EGF in response to N. meningitidis and depletion of HB-EGF results in reduced numbers of invasive bacteria [35], indicating that both pathways are relevant to N. meningitidis invasion into BECs.Here, we provided evidence that demonstrates the capacity of S1P to stimulate EGFR transactivation through the S1PR2 receptor in BECs infected with N. meningitidis.By blocking S1PR2 signaling, the highly selective S1PR2 antagonist JTE-013 attenuated EGFR phosphorylation at Tyr-845, indicating that transactivation of EGFR is mediated by Src signaling.
EGFR is considered an important receptor manipulated by diverse pathogens for their survival in the host [58][59][60].It was recently shown that transactivated EGFR recruits α-actinin-4, an actin-cross-linking protein, from intracellular F-actin fibers, thereby disrupting the cytoskeleton to facilitate E. coli invasion of BECs [61].Of note, α-actinin-4 recruitment is also relevant for N. meningitidis adhesion, as α-actinin-4 binds directly to the cytosolic tail of CD147 and directs the assembly of CD147-β2AR complexes in highly ordered clusters at bacterial adhesion sites [32].Whether α-actinin-4 recruitment may also play a role in S1PR2-EGFRmediated uptake of N. meningitidis requires further investigation.
S1PR2 signaling via Rho/ROCK can induce vascular permeability through stress fiber formation and disruption of adherence junctions.ROCK can directly destabilize tight junctions by phosphorylation of tight junction proteins occludin and claudin-5 [62].In addition, ROCK-mediated phosphorylation of the myosin-binding subunit of myosin-light chain phosphatase (MCLP) inhibits MCLP activity and hence leads to cellular contraction through reduced MCL dephosphorylation [63].Moreover, ROCK can phosphorylate MCL and therefore directly contributes to cellular contraction [64].Overexpression of S1PR2 in HUVECs completely abolishes S1P-induced cortical actin formation, whereas S1PR2 inhibition with JTE-013 prior to S1P treatment enhances cortical actin formation, indicating a fine-tuned regulation of cortical actin stabilization through different S1P receptors [65].It should be mentioned that S1PR2 is the only S1PR that can activate both the small GTPase Rho and ERM (ezrin/radixin/moesin), a family of adaptor molecules linking the cortical actin cytoskeleton to the plasma membrane [66].This might be relevant to its unique action as it has been suggested that pilus-mediated adhesion of N. meningitidis recruit ERM and together with Rho are critical for the subsequent cytoskeletal modifications responsible for the formation of protrusions and bacterial internalization [67].
Surprisingly, targeting S1PR3 by RNAi-mediated silencing or targeting S1PR1 with selective antagonist W146 increased bacterial uptake into BECs.No effect on bacterial uptake was however observed after treatment with S1PR1 siRNA, possibly due to a low turn-over rate or remaining expression of S1PR1.Treatment with S1PR1-specific agonist CYM5442 as well as S1PR3-specific agonist CYM5541, however, resulted in reduced bacterial uptake suggesting that binding of S1P to these receptors reduces numbers of invasive bacteria.Here, we showed that EGFR phosphorylation was only slightly affected by modulation of S1PR1/3 signaling, suggesting that these receptors influence bacterial entry via a different mechanism.Interestingly, both S1PR1 and S1PR3 increase the stability of the endothelial barrier by strengthening the cortical ring through Rac-dependent mechanisms.S1P exerts its barrier-strengthening effect via Rac-dependent translocation of cortactin to the cell periphery, but also requires the interaction of cortactin with endothelial myosin light chain kinase MLCK [68].While S1P signaling via S1PR1 enhances α-actinin 1/4 recruitment via Rac/PI3K/Tiam1 [69], activation S1PR1+3 leads to the inactivation of cofilin via Rac/PAK/LIMK resulting in reduced severing and overall stabilization of the of the cortical actin network [70].Conversely, constant actin disassembly by cofilin is essential for the formation of membrane ruffles, which form the phagocytic cup as a pre-requisite for cellular uptake as shown for example for Listeria monocytogenes [71].It is conceivable that N. meningitidis entry into hCMEC/D3s was inhibited by a similar mechanism, and future studies will address the question of whether activation of S1PR1+3 stabilizes the cortical cytoskeleton of BECs and inhibits the formation of membrane protrusions to prevent N. meningitidis uptake.
Several pathogenic bacteria can bind to and cross the endothelial cells of the BBB and cause meningitis [72,73].Consistent with our study, meningitic E. coli have also been shown to modulate sphingolipid homeostasis in BECs [74].In contrast to N. meningitidis, meningitic E. coli induce phosphorylation of SphK2 releasing S1P, thus resulting in transactivation of EGFR downstream of S1PR2 important for E. coli penetration of the BBB [74].
S1P receptors are variably coupled to three heterotrimeric G proteins, G αi , G 12/13 and G q .While S1PR1 couples exclusively to the G protein G αi , S1PR2/3 couple to all three G proteins, G αi , G 12/13 and G q [75].In vitro studies indicate that S1PR1-G αi coupling directly acts on Rac, leading to rearrangement of the cytoskeleton of endothelial cells and formation of a cortical ring, resulting in cellular spreading and flattening [70].In addition, by binding to S1PR1, S1P increases translocation of the adherence junction component VE-cadherin to cell-cell contact regions to further stabilize endothelial barrier integrity [76].Opposingly, S1PR2 via coupling to G 12/13 , leads to endothelial leakage through Rho-mediated Rac inhibition [65].The descripted function of S1PR3 is more controversial.While S1PR3 knockdown mice are protected against LPS-induced disruption of the lung endothelial barrier [77], in vitro experiments show that S1PR3 along-side with S1PR1 promotes endothelial barrier function via G αi signaling [70].
The functional balance between signal transduction after activation of S1PR1/3 and S1PR2, respectively, at the BBB is well established, and confers to protection of barrier integrity via regulation of tight junction proteins or cytoskeletal rearrangement [22].While plasma S1P levels are controlled systemically, it is possible that S1P levels at the BBB are controlled through local changes of S1P modulating enzymes and therefore S1P receptor signaling at the BBB could be controlled by endothelial cells in an autocrine manner by S1P release and S1P receptor expression [78,79].Future studies are needed to determine the impact of S1P and S1PRs on the BBB disruption in response to N. meningitidis infection.
Despite advancements in antibacterial therapy, bacterial meningitis remains one of the most important infectious diseases worldwide [80].The most common pathogens in children and adults include N. meningitidis, which is associated with high morbidity and mortality.Here we show that N. meningitidis modulates the sphingolipid content on BECs to efficiently invade BECs.Our study provides new insights into the underlying mechanisms-which involves SphK1 activation and a novel signaling mechanism, that mediates signals between three individual ligand-receptor system-and thus a basis for the development of a potential adjuvant strategy to combat bacterial meningitis caused by this pathogen.

Cell line and culture conditions
hCMEC/D3 cells (derived from human temporal lobe microvessels isolated from tissue removed during operations for control of epilepsy [81], Merck) were cultured in Endo-GRO-MV Complete Culture Medium (Merck) supplemented with 20 ng/L hBFGF (Merck) on 150 μg/mL collagen 1 (Gibco) coated T75 flasks at 37˚C and 5% CO 2 .For sub-cultivation and infection experiments, hCMEC/D3 were detached using Trypsin/EDTA solution (Ther-moFisher) and seeded in optimum density.Cells from passage 10 to 25 were used for experiments.hCMEC/D3 cell line was authenticated by ATCC cell line authentication service using short tandem repeat analysis [82].Cells were regularly checked for mycoplasma contaminations using a commercially available kit (PanReac AppliChem).

Bacterial strains and growth conditions
Neisseria meningitidis serogroup B strain MC58 (sequence type [ST74]/ST-32 clonal complex (CC)) [83] is a clinical isolate from UK and was kindly provided by E.R. Moxon.
To generate an isogenic non piliated mutant of strain MC58, the pilE gene was amplified using oligonucleotides pilE_XbaI_fwd and pilE_XhoI_rev, cleaved with XbaI and XhoI, respectively, and cloned into the pTL1 vector [86].The pTL1 plasmid, harboring the pilE gene fragment was then transformed into TOP 10 chemically competent E. coli cells.After confirmation of positive clones by nucleotide sequencing, an inverse PCR was performed using olidonucleotides pilE_AvrII_inv_fwd and pilE_AvrII_inv_rev harboring an AvrII restriction site, the resulting construct was cutted with AvrII and a spectinomycin-resistance cassette was inserted via the AvrII sites by replacing about 400 bp of the coding sequence.The resultant construct was amplified in E. coli TOP10 before transformation of the plasmid into N. meningitidis strain MC58.N. meningitidis MC58 pilT mutant, which lacks the pilus retraction machinery, was obtained with the same cloning strategy using pilT_XbaI_fwd, pilT_X-hoI_rev, pilT_AvrII_inv_fwd and pilT_AvrII_inv_rev, respectively.All primer sequences are listed in Table 1.Mutant strains were confirmed by PCR, sequencing, and Western blot analysis.

Inhibitors, agonists, and antagonists
All Inhibitors, agonists and antagonists were purchased from commercial sources and dissolved in DMSO (Carl Roth) (final DMSO concentration of 0.1%).For infection experiments, PF543 (Echelon Biosciences), FTY-720-phosphate (Echelon Biosciences), W146 (Merck) and JTE-013 (Merck) were dissolved in DMSO to yield a stock solution of 10 mM.FTY-720 (Echelon Biosciences), CYM5520 (Cayman Chemicals) and SLM-6031434 (Tocris) were dissolved in DMSO to yield a stock solution of 1 mM.K145 (Merck) was dissolved to a stock solution of 2.5 mM, CYM5442 (Cayman Chemicals) to a stock solution of 0.5 mM and CYM5541 (Cayman Chemicals) to a stock solution of 7 mM, all in DMSO.

PLOS PATHOGENS
SphK1/S1PR axis is essential for uptake of N. meningitidis in BECs

Cell viability assay
The cytotoxic effects of all inhibitors, agonists, antagonists, or siRNA against SphKs used in this study were determined via flow cytometric measurement of propidium iodide positive cells.In brief, hCMEC/D3s were seeded on 24-well plates, grown until confluence and treated with inhibitors, agonists, antagonists, or solvent as negative control for 8.5 h.For cytotoxicity measurements of RNAi experiments, cells were transfected with indicated siRNA concentrations or scrambled siRNA for indicated periods of time and incubated in EndoGRO medium supplemented with hBFGF for 8.5 h.Cells were washed with PBS, trypsinized for 5 min and centrifuged at 1500 g for 5 min.Cells were stained for 10 min on ice with 150 μL propidium iodide (10 μg/mL, Merck) / ribonuclease A (RNAse A) (2.5 mg/mL, Merck) solution and afterwards diluted by adding 150 μL of PBS.hCMEC/D3s treated with 0.25% Triton X-100 (Carl Roth) served as positive control.Flow cytometric measurement of at least 10,000 gated events per sample was conducted using MACS Quant10 (Miltenyi).The percentage of propidium iodide negative cells was determined from 10,000 events measured after exclusion of cellular debris and doublets via forward and sideward scatter using FlowJo analysis software.

Growth curves
Bacteria from liquid cultures were centrifuged at 3320 x g for 5 min, resuspended in cell culture medium and optical density at 600 nm (OD 600 ) was measured to calculate the bacterial inoculum (OD 600 = 1 ≙ 1*10^9 bacteria/mL).1*10^8 bacteria were diluted in 10 mL Endo-GRO medium supplemented with hBFGF containing indicated concentrations of inhibitors, agonist, or antagonist (S5 and S6 Figs) or DMSO as negative control and incubated with shaking (200 rpm) at 37˚C.OD 600 were recorded every 30 min for 180 min.

Infection assays
Bacteria were seeded a day before the infection assay on Columbia sheep blood agar plates or GC agar plates supplemented with spectinomycin from a stock conserved at -80˚C.Bacteria were transferred to 10 mL of PPM+ for subsequent growth under agitation for 90 min.Bacteria from liquid cultures were centrifuged at 3320 xg for 5 min, resuspended in cell culture medium and CFUs were evaluated by OD 600 measurements and diluted to provide a multiplicity of infection (MOI) of 100.hCMEC/D3 cells were seeded onto different well plate sizes coated with 150 μg/mL collagen 1 and grown until confluence.EndoGRO medium supplemented with hBFGF was replaced 24h prior to infection with serum-starvation medium (sterile-filtered EndoGRO Basal medium (Merck) + 0.1% BSA (PanReac AppliChem)) to synchronize the cells.On the day of infection, cells were washed once with PBS, and EndoGRO medium supplemented with hBFGF containing the respective inhibitors, agonists, or antagonists or 0.1% DMSO as negative control was added 30 min prior to infection.Gentamicin protection assays were then performed as previously described [6].Numbers of adherent and invasive bacteria were normalized to the controls of the individual experiments.

LC-MS/MS measurement
hCMEC/D3 cells were grown until confluency in 6-well plates, reaching approximately 1.5*10^6 cells/well at the day of infection, and were infected as described using an MOI of 100.At indicated time points, supernatants (1 mL) were collected and mixed with 3 mL 1-butanol (Carl Roth) for lipid extraction as described [87].In parallel, cells were washed with PBS, trypsinized for 5 min at 37˚C and collected in a reaction tube.After centrifugation for 5 min at 2000 g, supernatant was discarded and cells were resuspended in 1.5 mL methanol/chloroform (2:1, v:v) (Carl Roth/Supelco) for lipid extraction as described [88].

RT-qPCR
hCMEC/D3s were infected as described above.Sample collection and RNA purification was performed according to NucleoSpin RNA purification protocol (Marcherey Nagel).In brief, 350 μL RA1 buffer containing 1% 2-Mercaptoethanol (Merck) were added, technical duplicates were pooled to ensure sufficient RNA yield.and frozen until RNA purification was carried out.RNA concentration and purity was measured with NanoDrop One (ThermoFisher).500 ng RNA were used for cDNA synthesis with LunaScript RT SuperMix (New England BioLabs) according to manufacturer's protocol.cDNA samples were diluted 1:10 in nuclease-free H 2 O and used for RT-qPCR with PowerUp SYBR Green master mix (ThermoFisher) and 200 nM of respective primers listed in Table 2. RT-qPCR was performed using StepOnePlus real-time PCR thermocycler (ThermoFisher) and analyzed using StepOnePlus software (ThermoFisher).Fold change gene expression compared to negative control was calculated in Microsoft Excel following the delta delta Ct method with 18S rRNA as reference gene.For all Primers used for RT-qPCR, annealing temperature was optimized and efficiency (E = 95-105%) and linearity (pearson R > 0.999) was validated.Specificity was controlled by melt curve analysis and agarose gel electrophoresis.

Preparation of pilus-enriched fractions
Pilus-enriched fractions (PeF) were prepared as previously described [7].Briefly, 50 plates of N. meningitidis 8013/12 or N. meningitidis 8013/12 pilE were incubated overnight.The next day, the total bacterial content was swapped and transferred to 60 mL ethanolamine solution

PLOS PATHOGENS
SphK1/S1PR axis is essential for uptake of N. meningitidis in BECs (150 mM, pH = 10.5) in PBS.The solution was vortexed for 2 min to shear off pilus proteins.Bacterial cells and smaller debris were removed by a first centrifugation step at 12,000 x g for 10 min followed by a second centrifugation step at 21,500 x g for 90 min.The supernatant was then transferred to an Erlenmeyer flask and 6.6 mL ammonium sulfate-saturated (0.7 g/mL, Merck) ethanolamine solution (150 mM in PBS, pH = 10.5, Carl Roth) was added and stirred for 30 min on a magnetic mixer.The resulting precipitate, containing the pilus proteins, was pelleted through centrifugation at 21,000 x g for 15 min.The supernatant was discarded, and the pellets were dissolved in 50 mL TBS (0.05 M TRIS, pH = 7.5, Carl Roth).The protein solution was transferred to a VivaSpin column (50 mL, molecular weight cut-off 5 kDa, Merck).The columns were centrifuged at 4000 x g until the volume of the upper fraction was reduced to 1 mL.Protein content was determined using Pierce BCA assay (ThermoFisher).Pilusenrichment and absence of Opc was verified by Western blotting.1 μg of purified proteins (PeF) or TBS as solvent control was used for treatment of cells.

Sphingosine kinase assay
Activity of sphingosine kinases was determined using a commercially available sphingosine kinase activity assay kit (Echelon Biosciences # K-3500).In brief, hCMEC/D3s were grown on 96-well plates, pre-treated and infected as described.At indicated time points, cells were washed once with PBS before adding 50 μL of the reaction buffer containing 0.1%

PLOS PATHOGENS
SphK1/S1PR axis is essential for uptake of N. meningitidis in BECs dithiothreitol (DTT).Cell lysis was achieved by three repeated freeze-thawn cycles.10 μL of cell lysate was mixed with 400 μM sphingosine (in 10 μL reaction buffer containing 0.1% DTT) and 20 μM ATP (in 20 μL containing 0.1% DTT) in 96-well plates.The plates were sealed and incubated for 2 h at 37˚C.After incubation, 40 μL of luciferase-based ATP detector was added for further 10 min to stop the kinase reaction.Kinase activity was tested using Spec-traMax iD3 ELISA reader (Molecular Devices).To test the specificity of the commercially available ATP depletion-based assay, we validated that baseline SphK activity was significantly reduced without addition of Sph as substrate (Fig 2H ) as well as after siRNA-based knockdown of SphK1 and SphK2 (S1A Fig) .For blocking the interaction between Tfp and the receptor CD147, 10 μg/mL of anti-CD147 (clone MEM-M6/6, antibody, that binds to the extracellular N-terminal domain, Biorad # MCA2882Z) or 10 μg/mL anti-CD147 (clone MEM-M6/1, antibody, that binds to the C-terminal Ig domains, Biorad # MCA1876) were used and added 30 min before treatment of cells with PeF.

Western blot analysis
For determination of EGFR protein level and phosphorylation, hCMEC/D3s were grown on 6-well plates in conditioned medium and left either uninfected, infected with N. meningitidis or treated with S1P or EGF as positive controls.S1P (Avanti Polar Lipids) was dissolved in methanol: water according to the manufacturer's protocol and diluted to the desired concentration of 125 μM in fat-free BSA (4 mg/mL in PBS).At indicated time points, supernatants were removed, cells were washed with 500 μL ice-cold PBS and harvested in 300 μL ice-cold modified RIPA2 buffer (50 mM TRIS-HCl (pH = 7.2), 150 mM NaCl (Carl Roth), 5 mM EDTA (PanReac), 1% Triton X-100, 24.1 mM Sodium deoxycholate (Merck), 0.1% SDS (Carl Roth), 50 mM Sodium fluoride (Merck), 1% protease and phosphatase inhibitor cocktail (ThermoFisher)).Cells were collected using a cell scraper, agitated on a rotary shaker for 45 min on ice and lysates were centrifuged at 12,000 x g for 20 min at 4˚C.Lysates were diluted in 2x SDS reducing sample buffer (140 mM TRIS-HCl (pH 6.8), 22.4% Glycerol (Carl Roth), 4.5% SDS, 6.9% 2-Mercaptoethanol, Bromophenol blue (Merck)).For the determination of EGFR expression and phosphorylation, lysates were incubated for 10 min at 95˚C.
For determination of SphK1/2 expression and phosphorylation, cell lysates were generated using the same protocol, however lysates were incubated for 10 min at 70˚C.Cytosolic and plasma membrane fractions were isolated using a plasma membrane protein isolation kit (Invent BioTechnologies).In brief, hCMEC/D3s from two 175cm 2 flasks (appr.5*10 7 cells) were scraped in ice-cold PBS and treated according to the manufacturer's protocol without the filtration step to prevent nucleic burst.Total protein content of cytosolic and plasma membrane fractions was measured by Pierce BCA Assay (ThermoFisher).Equal protein amounts were loaded on SDS Gels.
Samples were probed on 12% SDS-PAGE gels and transferred to nitrocellulose membranes (Merck).Then, membranes were blocked with 5% BSA in TBS-T 1 h at RT and probed with phospho-specific antibodies at 4˚C o/n, washed three times with TBS-T.Anti-rabbit or antimouse secondary HPR-Antibody (JacksonResearch Labs) were used to incubate the membranes for 1h at RT and membranes were washed 3 times with TBS-T.Finally, the protein bands were detected with ECL substrate (Bio-Rad) and imaged using a chemiluminescence detector (Bio-Rad).Blots were subsequently stripped with mild stripping buffer (200 mM Glycine (Carl Roth), 0.1% SDS, 1% Tween20, pH = 2.2) and then reprobed with anti-SphK1 (Cell Signalling), anti-SphK2 (Proteintech), or anti-EGFR (Santa Cruz), washed and incubated with HRP-conjugated secondary antibody.Primary and secondary antibodies with recommended

PLOS PATHOGENS
SphK1/S1PR axis is essential for uptake of N. meningitidis in BECs dilution are given in Table 3. GAPDH and β-Actin (Proteintech) served as protein loading controls.The blot images were quantified using FIJI ImageJ analysis software, and the values for each phosphorylated target protein were normalized to target protein expression.The values for phosphorylation levels in infected cells were calculated relative to the values of phosphorylation levels in uninfected control cells.

Statistics
Data represent mean ± standard deviation.Statistical significances of at least 3 independent experiments were calculated using GraphPad Prism version 6.01 (GraphPad Software Inc.).Pairwise comparisons were carried out using two-tailed students t-test.If multiple t-tests were

PLOS PATHOGENS
SphK1/S1PR axis is essential for uptake of N. meningitidis in BECs employed, p-values were adjusted with Holm-Sidak correction.Comparisons with more than two groups were carried out using one-or two-way ANOVA followed by Dunnett's post-hoc test to determine significance compared to control group.P values lower than 0.05 were considered significant.Experiment replicates and statistical analyses are indicated in the figure legends.Information about data normalization can be found in the figure legends.

Software
Calculations were performed and graphs were created in Microsoft Excel version 2303 and GraphPad Prism version 6.01.Western Blot band intensity quantification was done using FIJI ImageJ version 1.52 [97].LC-MS/MS raw data were analyzed with MassHunter Quantitative Analysis 10.1 (Agilent Technolgies).

Fig 1 ,
Heatmaps).Cellular Sphingomyelin (SM) levels were non-significantly reduced over 8h infection time course (Fig 1A, Heatmap), most likely due to the turnover to Cer, whereas the extracellular SM levels were only marginally increased (Fig 1B, Heatmap).Cellular Sphingosine (Sph) levels increased in response to infection with N. meningitidis MC58, peaking 2h post-infection (p.i.) (Fig 1A), whereas the Sph levels in the supernatant did not change (Fig 1B).Importantly, cellular S1P significantly increased reaching a peak 4h p.i. and returned to basal levels by 8h (Fig 1A).Moreover, S1P levels increased continuously in the supernatant, reaching 240% of basal S1P levels over the 8h infection time course (Fig 1B).

Fig 1 .Fig 2 .
Fig 1. N. meningitidis induces elevation of sphingosine 1-phosphate levels in hCMEC/D3s.(A+B) hCMEC/D3 were infected with N. meningitidis strain MC58 at a MOI of 100, and sphingolipid levels were quantified at indicated time points by LC-MS/MS in cell lysates (cellular, A) or in supernatants (extracellular, B).Ceramide (Cer), Sphingosine (Sph), Sphingomyelin (SM) and Sphingosine 1-phosphate (S1P).Heatmap representing the log2 fold-change values of measured sphingolipid species between infected and uninfected control cells (control).Graphs/Diagrams show percentage changes of Cer (upper panel), Sph (middle panel) and S1P (lower panel) relative to uninfected control cells.The values of fold changes are indicated by the coloured scale, with deep red indicating a value>1 and blue indicating a value <1.Data are the means of four biological replicates ± SD. ** p < 0.01, *** p < 0.001, **** p < 0.0001 as calculated with one-way ANOVA followed by Dunnett's post-hoc test compared to uninfected control cells.https://doi.org/10.1371/journal.ppat.1011842.g001

Fig 4 .
Fig 4. N. meningitidis induces S1PR2-mediated EGFR activation to enable invasion into hCMEC/D3s.(A) Relative mRNA expression of S1P receptors S1PR1, S1PR2, S1PR3, S1PR4 and S1PR5 in hCMEC/D3s determined by quantitative RT-PCR.Data are normalized to 18S rRNA relative to S1PR5 mRNA expression.(B) S1PR1 expression, (C) S1PR2 expression, (D) S1PR3 expression in hCMEC/D3s infected with N. meningitidis MC58 for 8h.mRNA expression levels were analyzed by qRT-PCR with normalization to 18S rRNA relative to uninfected control cells.Data are shown as mean ± SD, n = 4 experiments measured in duplicates.Comparison to control for each time point using multiple t-test and p-value adjustment with and 1 μM prevented significant EGFR phosphorylation 6h p.i., without affecting N. meningitidis adhesion (S4C Fig).JTE-013 treatment was sufficient to inhibit N. meningitidis MC58 uptake by hCMEC/D3 until 6h but not 8h infection (Fig 4H), possibly due to increasing expression of S1PR2 (Fig 4C).Therefore, the contribution of S1PR2 to N. meningitidis uptake by hCMEC/D3 was confirmed by siRNA knockdown of S1PR2 that reduced expression of S1PR2 by 80% without affecting expression of S1PR1 or S1PR3 (S2B Fig).Downregulation of S1PR2 significantly reduced numbers of recovered intracellular N. meningitidis MC58 even 8h p.i. (Fig 4I).In support of these findings, pre-treatment with selective S1PR2 agonist CYM5520[46] significantly increased N. meningitidis uptake 4h p.i. (Fig 4J), indicating that S1PR2 activation directly precedes and enhances N. meningitidis endocytosis.Taken together, our results suggest that an increased release of S1P, together with enhanced expression of S1PR2 triggered by N. meningitidis, leads to transactivation of EGFR, which allows the pathogen to invade BECs.
) and mRNA expression of S1PR3 by 71% (S2C Fig), respectively.Holm-Sidak correction.** p < 0.01 vs. respective control.(E) Determination of EGFR protein levels and phosphorylation at Tyr-845 (p-EGFR) in hCMEC/D3 cell lysates by Western blot analyses.hCMEC/D3s were stimulated with 1 μM S1P or 25 ng/mL EGF for 15 min, or additionally infected with N. meningitidis MC58.Cells were lysed with RIPA buffer, and cell lysate proteins were resolved with SDS-PAGE.Tyrosine phosphorylation of EGFR was analyzed using phospho-specific EGFR antibody (Tyr-845).Stripped membranes were blotted with EGFR antibody.A representative Western blot is shown.Densitometric analyses show data of two independent experiments.(F) Determination of EGFR expression and phosphorylation at Tyr-845 (p-EGFR) in hCMEC/D3s infected for 8h by Western blot and assessment by densiometric analyses.Data represent means ± SD, n = 3. One-way ANOVA followed by Dunnett's post-hoc test, ***p<0.001.(G) Assessment of pharmacological inhibition of S1PR2 on EGFR protein levels and phosphorylation at Tyr-845.hCMEC/D3s were either pretreated with S1PR2/4 antagonist JTE-013 or solvent control (DMSO) and infected with N. meningitidis MC58 or left uninfected.Cell lysates were analyzed by Western blot and subsequent densiometric analysis.Data represent means ± SD, n = 4. One-way ANOVA followed by Dunnett's post-hoc test, *p<0.05,**p<0.01.(H, I, J) Assessment of siRNA targeted knockdown of S1PR2 or pharmacological modulation of S1PR2 on bacterial invasion into BECs.hCMEC/D3 were pre-treated with (H) JTE-013 or (J) CYM5520 for 30 min or (I) with siRNA specific for S1PR2 for 72h.Cells were subsequently infected with N. meningitidis MC58 for 8h.Numbers of viable intracellular bacteria were determined by gentamicin protection assay and normalized to untreated control cells (DMSO control or scrambled siRNA for each time point).Data are shown as mean ± SD; n � 4. Unpaired students t-test (I) or multiple t-test and p-value adjustment with Holm-Sidak correction.* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.Absence of cytotoxicity, knockdown efficiency and bacterial growth were verified as shown in S2 and S4 Figs.https://doi.org/10.1371/journal.ppat.1011842.g004

Table 1 . Oligonucleotides used in this study.
For work-up of both, supernatants and cells, the extraction solvent contained the internal standards d 7 -sphingosine (d 7 -Sph), d 7 -sphingosine 1-phosphate (d 7 -S1P), 17:0 Cer and d 31 -16:0 SM (all Avanti Polar Lipids).Chromatographic separations were achieved on a 1290 Infinity II HPLC (Agilent Technologies) equipped with a Poroshell 120 EC-C8 column (3.0 × 150 mm, 2.7 μm; SM, m/z 813.7 !184.1 (86.1) for 24:1 SM, and m/z 815.7 !184.1 (86.1) for 24:0 SM.Peak areas of Cer and SM subspecies, as determined with MassHunter Quantitative Analysis software (version 10.1, Agilent Technologies), were normalized to those of the internal standards (17:0 Cer or d 31 -16:0 SM) followed by external calibration in the range of 1 fmol to 50 pmol on column.Sph and S1P were directly quantified via their deuterated internal standards d 7 -Sph (0.25 pmol on column) and d 7 -S1P (0.125 pmol on column).Changes in sphingolipid content were calculated by normalizing the determined concentrations to the concentrations of uninfected cells for each experiment and for S1P to the respective time point.