Background
Lysophospholipids are metabolites of glycerophospholipids and sphingolipids that are commonly found as lipid constituents of cell membranes. Besides being a structural membrane component, certain members of the lysophospholipid family have considerable cell signaling properties [
1]. Among these, sphingosine-1-phosphate (S1P) and lysophosphatidic acid (LPA) and their signaling pathways have been best characterized. Lysophospholipid signaling appears to have a large range of physiological and pathophysiological functions [
2,
3] in the adult organism and also plays important roles during embryonic development, especially in the development of the nervous system and vascular development [
4‐
6].
S1P acts as a ligand for five G-protein-coupled receptors [
7‐
11], generally referred to as S1P
1–5. Since S1P receptors couple to a variety of G-proteins, stimulation of S1P receptors can impact several signal transduction pathways and cellular processes. In contrast to S1P
2–5, S1P
1 is thought to exclusively couple with G
i/o, affecting cyclic adenosine monophosphate (cAMP) levels and Ca
2+ mobilization, among others [
10,
12].
Fingolimod (also named FTY720) is a first-in-class S1P receptor agonist that has been approved for the treatment of remitting-relapsing multiple sclerosis due to its immunomodulatory properties [
13]. Structurally, a sphingosine analogue, fingolimod becomes phosphorylated by sphingosine kinases [
14] and binds to four of the five S1P receptors [
15], except for S1P
2. Its immunomodulatory effect is presumably based on its “functionally antagonistic” effect on the S1P
1 receptor expressed on lymphocytes, causing S1P
1 receptor internalization and subsequently abrogation of S1P
1-mediated signaling. It thereby prevents lymphocyte egress [
16,
17]. In addition, phosphorylated fingolimod (fingolimod-P) has been reported to act as an inhibitor of autotaxin, an ectonucleotide pyrophosphatase/phosphodiesterase with lysophospholipase D activity that generates LPA from more complex lysophospholipids [
18]. LPA has been implicated in dorsal root demyelination after partial sciatic nerve injury, most likely by activation of LPA
1 receptor signaling [
19], and demyelination has been found to be significantly reduced in heterozygous autotaxin knockout mice [
20]. Furthermore, traumatic brain injury has been reported to induce significant increases of various LPA species in cerebrospinal fluid already 3 hours after injury in mice, and administration of LPA-antibody improves brain tissue damage outcomes [
21]. Additionally, recent publications point to a significant role of LPA and LPA
1 receptor signaling in the pathophysiology of spinal cord injuries [
22,
23].
As S1P receptors are known to be widely expressed in the nervous system [
24], there is an emerging body of experimental evidence pointing to direct effects of fingolimod on cells of the central nervous system [
25], including neuroprotection from excitotoxic death and the promotion of remyelination. Additionally, fingolimod has been reported to promote functional recovery in spinal cord injury models [
26,
27]. However, the effects of fingolimod on cells of the peripheral nerve in vivo have not been investigated so far. To address the question whether fingolimod may promote peripheral axon regeneration or remyelination and to differentiate between potential direct PNS-specific and its established immunomodulatory effects, we performed sciatic nerve crush in wildtype C57BL/6 as well as in two immunodeficient mouse strains:
Rag1
−/−
mice lacking mature B- and T-lymphocytes and in athymic
Foxn1
−/−
mice devoid of T-lymphocytes. To further explore the role of LPA during sciatic nerve de- and regeneration as well as to distinguish the significance of S1P receptor modulation from LPA mediated effects, we treated animals with the autotaxin inhibitor PF-8380 [
28] and assessed the effect of fingolimod on LPA formation after injury in sciatic nerve around the crush site.
Methods
Animals
Rag1
−/−
mice (B6.129S7-Rag1
tm1Mom
/J) were obtained from Jackson Laboratories (Bar Harbor, ME, USA), Foxn1
−/−
mice (B6.Cg/NTac-Foxn1
nu
) from Taconic (Hudson, NY, USA), and control C57BL/6 mice from Janvier Labs (Le Genest-Saint-Isle, France). Animal use and experiments were approved by local authorities (LANUV North Rhine-Westphalia, Germany).
Sciatic nerve crush
Male, age-matched (3–4 months) wildtype C57BL/6, Rag1
−/−
, and Foxn1
−/−
mice were anesthetized for surgery via intraperitoneal injection of a mixture of xylazine (Rompun; Bayer, Leverkusen, Germany) (10 mg/kg) and ketamine (Actavis, Munich, Germany) (100 mg/kg) and placed on a heating plate (37 °C) to maintain constant body temperature. The fur of the lower back was removed with an electric razor, and the skin was disinfected using 70 % ethanol. All instruments were sterilized. A small incision (1 cm) was made in the skin above the right hindlimb between the mm. gluteus maximus and biceps femoris. Opening the facial plane between both muscles revealed the sciatic nerve which was carefully lifted using bent forceps and crushed right before its distal branches using a non-serrated clamp at maximum intensity for 30 s. The nerve was replaced under the muscle, and the incision was closed using non-absorbable suture material. The contralateral nerve was left intact to serve as control.
Administration of fingolimod and PF-8380
Mice received non-phosphorylated fingolimod (FTY720, Cayman Europe, Tallinn, Estonia) dissolved in a solution of 10 % DMSO in PBS (Sigma-Aldrich, Munich, Germany) via intraperitoneal injection at a concentration of 1 mg/kg once daily over the course of 16 days, starting 2 days before crush until 14 days post-crush. PF-8380 (Sigma-Aldrich) was dissolved in DMSO and administered at a concentration of 10 mg/kg via intraperitoneal injection once daily, as well starting 2 days before until 14 days post-crush. Controls received an equal volume of solvent.
Clinical assessment of nerve functionality by walking track analysis
Nerve functionality was assessed 2 days before as well as 7 and 14 days post-crush by walking track analysis and calculation of the sciatic functional index (SFI) as described elsewhere [
29].
Electrophysiology
Electrophysiology was essentially performed as described previously [
30]. Nerve conduction velocities and compound muscle action potentials were determined at 14 days post-crush. Mice were anesthetized with a mixture of ketamine (100 mg/kg) and xylazine (10 mg/kg) and immediately placed on a heating plate (37 °C) to maintain constant body temperature. Stimulation of the sciatic nerve was performed by repetitively generated single pulses using monopolar 30 G needle electrodes until supramaximal stimulation was achieved. Compound muscle action potential was recorded at the plantar foot muscle with a needle electrode using a portable electrodiagnostic system (KeyPoint 4, Medtronic, Meerbusch, Germany). Nerve conduction velocity was calculated from the distance and the motor latency differences between proximal and distal stimulations.
Tissue preparation
Following electrophysiology, mice were sacrificed via cervical dislocation. Sciatic nerves were carefully removed by only handling the most proximal end with forceps and cutting the nerve at its most distal end using scissors. For immunohistochemical applications, the nerves were immediately dipped in an isopentane bath immersed in liquid nitrogen for approximately 5 s. Then frozen nerves were placed in suitable cryomolds, covered with a cryo-embedding compound and placed on dry ice. Embedded nerves were stored at −80 °C. Longitudinal sections of 7-μm thickness were prepared in a cryostat chamber, and slides were air-dried for at least 1 hour before further processing or stored at −20 °C.
Antibodies
The following antibodies were used and diluted in Antibody Diluent (Dako, Hamburg, Germany) as indicated: Rabbit anti-cAMP polyclonal antibody—1:50 (Merck-Millipore, Darmstadt, Germany); biotinylated goat anti-rabbit IgG—1:200 (Vector Laboratories, Peterborough, UK); DyLight 594 Streptavidin—1:200 (Vector Laboratories); rabbit anti-neurofilament L (NF-L)—1:1000 (Merck-Millipore); and Alexa Fluor 488 Goat Anti-Rabbit IgG—1:200 (Life Technologies).
Immunofluorescence
Sections were post-fixed in 4 % paraformaldehyde for 20 min. After fixation, slides were washed 5 min in PBS and twice for 5 min in PBT (PBS + 0.1 % Triton X-100). Slides were incubated with blocking solution (10 % normal goat serum (v/v) in PBT) for 30 min at room temperature. For the detection of cAMP, primary antibody was applied and slides were incubated at 4 °C for 16 h. Slides were washed twice for 5 min in PBT, and biotinylated secondary antibody was applied and incubated at room temperature for 1 hour. Slides were washed twice for 5 min in PBT, and DyLight594-conjugated streptavidin was applied. If desired, co-staining with NF-L was performed. Slides were washed 2 × 5 min in PBT and incubated with anti-NF-L antibody at room temperature for 1 hour. Slides were washed twice for 5 min and incubated with Alexa Fluor 488-conjugated secondary antibody. Slides were washed 5 min in PBT and 5 min in PBS and mounted with Vectashield Hardset mounting medium containing DAPI (Vector Laboratories).
Preparation of semi-thin sections
Nerves were fixed in 0.1 M cacodylate buffer containing 2.5 % glutaraldehyde and kept at 4 °C overnight. The fixative was discarded and replaced by washing buffer (0.1 M cacodylate + 3 % sucrose). Nerves were washed for 4 days at 4 °C. Washing was followed by incubation in an osmium tetroxide reagent for 3 h. Osmium tetroxide reagent was composed of one part 5 % potassium dichromate solution (pH 7.4), one part 3.4 % NaCl solution, and two parts 2 % osmium tetroxide solution (Sigma-Aldrich). Afterwards, samples were briefly washed in 0.1 M cacodylate buffer.
Samples were dehydrated in an ascending ethanol series (70 %; 96 %; ≥99.8 % undenatured ethanol) for 1 hour each. Following dehydration, samples were incubated in 250 μl propylene oxide (Sigma-Aldrich) in tightly closed containers for 1 hour at room temperature, then 1 hour in a 1:1 mixture of propylene oxide/epon (epoxy embedding medium kit; Sigma-Aldrich), and finally kept at 4 °C in epon only overnight. Samples were placed in silicone molds and covered with epon embedding mixture. Embedded samples were incubated at 37 °C for 6 h, at 47 °C for 15 h, and finally at 60 °C for 28 h until epon was completely hardened. Transverse sections were prepared at a thickness of 1 μm at a Reichert-Jung Ultracut Microtome and immediately stained with toluidine blue (1 % toluidine blue (w/v) dissolved in a 1 % disodium tetraborate (w/v) solution), washed in distilled H2O (approximately 10 ml) containing 1–2 drops of acid ethanol (0.01 % HCl in absolute ethanol), placed on a microscope slide, dried on a heating plate, and mounted with Roti Histokitt II (Roth).
Quantification of LPA by liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS)
Sciatic nerve homogenates were extracted using 800 μl of a mixture methanol:chloroform (2:1) (both VWR International, Darmstadt, Germany) after addition of 20 μl LPA 17:1 (1 μg/ml) (Avanti Polar Lipids, Alabaster, AL, USA) as internal standard (IS), 50 μl deionized water (VWR), and 40 μl 6 M HCl (Merck, Darmstadt, Germany). Samples were vortexed for 5 min and incubated 20 min at -20 °C. After centrifugation (10 min, 14,000×g), the upper phases were transferred to new sample tubes and extracted with 200 μl chloroform (Merck) and 250 μl deionized water. Samples were vortexed rigorously and centrifuged for 10 min at 14,000×g. The lower organic phases were dried under nitrogen gas flow at 60 °C. The dried residues were dissolved in 50 μl LC-MS grade methanol (mobile phase B). Analysis was performed on a Waters/Acquity ultra performance LC-MS system controlled with MassLynx software. Analysis was operated with electrospray ionization (ESI) probe in negative ESI employing multiple reaction monitoring (mrm) mode. LPA species for the establishment of mrm experiments were obtained from Avanti Polar Lipids. The following transitions were used: m/z 435 [M-H-] to 153 (LPA 18:1), m/z 409 [M-H-] to 153 (LPA 16:0), m/z 437 [M-H-] to 153 (LPA 18:0), and m/z 421 [M-H-] to 153 for the internal standard (LPA 17:1). Collision energy (ce) and cone voltage (cv) were both set to 30 V for all mrm experiments. The column used for chromatography was a Thermo Scientific Hypersil Gold, 1.9 μm, 50 × 2.1 mm column. A total flow rate of 0.4 ml/min was applied with an injection volume of 10 μl. The mobile phases consisted of 5 mM ammonium formate (Sigma-Aldrich) buffered water (A) and methanol (B), both containing 0.1 % formic acid (Sigma-Aldrich). Gradient elution was performed starting at 50 % mobile phase B applying isocratic conditions for 2.5 min and then progressing to 100 % B over 3 min. 100 % B was held up for 2 min. Within 1 min, the gradient increased linear to initial conditions and was held up to 10 min for re-equilibration. Column temperature was adjusted at room temperature (25 °C ± 5 °C). The autosampler operated at 10 °C. Run time was set to 10 min. Retention times of LPA species were 2.99 min (LPA 18:1), 2.95 min (LPA 18:0), 3.10 min (LPA 16:0), and 2.17 min (LPA 17:1 IS), respectively. Data analysis was performed using peak-area-ratios of analytes to internal standard.
Determination of nerve cytokine levels by ELISA
For the detection of cytokines, commercially available ELISA kits (Duo Set mouse TNF-α, R&D Systems, Minneapolis, MN, USA; Mouse IL-10 ELISA MAX Standard Set, Biolegend, San Diego, CA, USA) were used essentially as instructed. Briefly, sciatic nerves were homogenized in 250 μl ice-cold PBS containing cOmplete™ Mini (Roche, Basel, Switzerland) proteinase inhibitor cocktail. Homogenates were centrifuged for 10 min at 11,000 rpm, and supernatants were transferred to new sample tubes, diluted 1:2 with assay buffer (PBS containing 1 % BSA and 0.1 % Tween20) and vortexed rigorously. Detection of cytokines was performed as recommended by the manufacturers. Sciatic nerve cytokine content was normalized to the respective protein content of the sample.
Image and data analysis
Analysis of images was performed using ImageJ (National Institutes of Health, Bethesda, MA, USA). Data analysis and compilation of graphs was performed using Microsoft (Redmond, WA, USA) Excel and GraphPad (La Jolla, CA, USA) Prism 5. Statistical analysis was done by Student’s t test, multiple comparisons were performed by one-way ANOVA followed by the Newman-Keuls post hoc test or Kruskal-Wallis test and Mann-Whitney U test for non-Gaussian distributions. Statistical significance is indicated by asterisks with P ≤ 0.05*, P ≤ 0.01**, and P ≤ 0.001***.
Discussion
In this work, we provide in vivo evidence for direct effects of the S1P receptor agonist fingolimod on the peripheral nervous system that may result in the promotion of peripheral nerve regeneration. Although certain positive effects of fingolimod-induced T-lymphocyte sequestration on nerve regeneration cannot be ruled out completely, our data strongly indicate that direct effects of fingolimod on neurons and Schwann cells are responsible for improved nerve regeneration. Fingolimod treatment was associated with a significant elevation of axonal cAMP, a crucial factor for central and peripheral axon regeneration [
31‐
34]. The observed increase in cAMP may most likely be the result of fingolimod-induced S1P
1 receptor internalization in cells of the peripheral nerve that may cause the abrogation of S1P
1-mediated inhibition of adenylate cyclase. It has been proposed that the S1P
1-G
i-adenylate cyclase system may become internalized as a ternary complex, leading to sustained inhibition of adenylate cyclase as long as the ligand fingolimod-P is bound [
39]. However, we propose differences between “short-term” (several hours) and “long-term” (several days to weeks) administration of fingolimod as, in contrast to inhibition of cAMP formation in cell culture experiments by sustained signaling from intracellular compartments [
40], our data demonstrate that long-term administration of fingolimod in vivo may result in an elevation of cAMP. The internalized S1P
1-G
i-adenylate cyclase system is likely to become degraded over time, and constantly high concentrations of fingolimod-P may prevent the localization of de novo synthesized S1P
1 to the plasma membrane. Hence, the pivotal difference in the long-term situation may be the spatial segregation of S1P
1 and adenylate cyclase, consequently allowing for an increased cAMP response in the regenerating nerve. Clearly, further studies are needed to corroborate this assumption.
In contrast to T-lymphocyte-deficient
Foxn1
−/−
mice which exhibited signs of significantly improved nerve regeneration under fingolimod treatment, B- and T-lymphocyte deficient
Rag1
−/−
mice did not show an improvement of nerve regeneration, neither with fingolimod nor vehicle treatment. Although one recent study has reported evidence for improved nerve regeneration after femoral nerve injury in
Rag2
−/−
mice which lack functional T- and B-lymphocytes as well [
41], others have found nerve regeneration and motor neuron survival to be reduced in
Rag2
−/−
mice [
42], which is in line with our observations in
Rag1
−/−
mice. The idea of an overall reduced regenerative capacity of
Rag1
−/−
mice is further supported by the finding that B-lymphocytes appear to be required for the production of autoantibodies against myelin debris containing inhibitors of axonal outgrowth, consequently allowing for normal axonal regeneration after nerve injury [
43]. Myelin-associated inhibitors of axonal outgrowth have been implicated in reducing neuronal cAMP in a G
i-dependent manner via activation of the Nogo receptor-p75 neurotrophin receptor complex [
44], providing a likely explanation for significantly reduced axonal cAMP in response to fingolimod treatment and, consistently, the impaired regenerative capacity observed in
Rag1
−/−
mice due to the lack of B-lymphocytes. Although fingolimod is known to also reduce the number of circulating B-lymphocytes, it has recently been shown to increase the proportion of regulatory B-lymphocytes producing IL-10. While we could not detect a significant elevation of IL-10 nor a decrease in TNF-α levels in the sciatic nerve of fingolimod-treated mice at 24-h post-crush (Fig.
4), an increase in IL-10 secreting B-lymphocytes may occur at later stages and could potentially contribute to accelerated nerve regeneration [
37,
45].
However, despite decreased cAMP,
Rag1
−/−
mice still showed significantly improved myelin thickness. Another member of the lysophospholipid family, LPA, was found to be significantly reduced in the crushed nerve in fingolimod-treated mice shortly after injury. Strikingly, inhibition of autotaxin by the specific inhibitor PF-8380 prevented the significant reduction of myelin thickness normally observed after peripheral nerve injury. Although LPA levels were not invariably and not significantly increased in control mice, it appears unlikely that a sole increase in LPA would account for demyelination, but rather an elevated expression of the LPA
1 receptor after injury. In this context, it has been reported that the LPA
1 receptor is upregulated within the first days after sciatic nerve injury in the distal nerve stump, and this upregulation is accompanied by an abrupt downregulation of the myelin gene
P0 [
46]. In line with this, a single intrathecal injection of LPA or nerve injury was found to cause a decrease in the expression of the myelin proteins PMP22 and MBP, but the demyelinating effect was largely abolished in LPA
1-null mice [
19]. Furthermore, antagonism of the LPA
1 receptor has recently been demonstrated to reduce demyelination after spinal cord injury [
21]. Therefore, improved myelination could be a consequence of both fingolimod as well as PF-8380-mediated inhibition of autotaxin, attenuating LPA
1 receptor activation via reduced LPA synthesis. Although autotaxin represents only one of multiple metabolic pathways to produce LPA, autotaxin appears to be one major contributor to LPA synthesis after injury, as it was previously shown that injury-induced dorsal root demyelination is significantly reduced in heterozygous autotaxin knockout mice [
19]. However, since PF-8380 treatment did not lead to signs of improved axonal regeneration, our results might suggest a dual mode of action for fingolimod in peripheral nerve regeneration—fingolimod may promote axonal regeneration by indirectly elevating cAMP formation as a consequence of S1P
1 receptor internalization and may prevent LPA-induced downregulation of myelin gene expression in the regenerating nerve. Further studies are warranted to corroborate this assumption.
Abbreviations
cAMP, cyclic adenosine monophosphate; Foxn1, forkhead box protein N1; IL-10, interleukin 10; LPA, lysophosphatidic acid; LPA1, lysophosphatidic acid receptor 1; NF-L, neurofilament light chain; Rag1, recombination activating gene 1; S1P, sphingosine-1-phosphate; S1P1, sphingosine-1-phosphate receptor 1; SFI, sciatic functional index; TNF-α, tumor necrosis factor alpha.
Acknowledgements
We like to thank Zippora Kohne for the excellent technical assistance.