Genetic deletion of Autotaxin from CD11b+ cells decreases the severity of experimental autoimmune encephalomyelitis

Autotaxin (ATX) is a secreted lysophospholipase D catalyzing the extracellular production of lysophosphatidic acid (LPA), a growth factor-like signaling lysophospholipid. ATX and LPA signaling have been incriminated in the pathogenesis of different chronic inflammatory diseases and various types of cancer. In this report, deregulated ATX and LPA levels were detected in the spinal cord and plasma of mice during the development of experimental autoimmune encephalomyelitis (EAE). Among the different sources of ATX expression in the inflamed spinal cord, F4/80+ CD11b+ cells, mostly activated macrophages and microglia, were found to express ATX, further suggesting an autocrine role for ATX/LPA in their activation, an EAE hallmark. Accordingly, ATX genetic deletion from CD11b+ cells attenuated the severity of EAE, thus proposing a pathogenic role for the ATX/LPA axis in neuroinflammatory disorders.


Introduction
Multiple sclerosis (MS) is the most common debilitating disorder of the central nervous system (CNS), imposing a substantial personal, and socioeconomic burden [1]. MS is a chronic inflammatory, autoimmune and neurodegenerative disease, characterized by loss of myelin sheaths and oligodendrocytes in the brain and spinal cord, resulting in aberrant axonal conduction leading to neurologic disabilities and, eventually, impaired mobility and cognition [1]. Despite the inherent limitations of modeling a human disease in experimental animals, valuable insights into MS pathophysiology were provided by the model of experimental autoimmune encephalomyelitis (EAE), that can be induced by immunizing animals against myelin antigens [2]. In this context, current research suggests that inflammation is a prominent feature of MS pathogenesis, initiated by autoreactive lymphocytes in the periphery and substantiated by leakage of the blood-brain barrier (BBB) [3]. Neuroinflammation leads to microglial and astrocyte activation, astrogliosis and eventually to demyelination, axonal or neuronal loss [3].
Microglia and macrophages are well recognized as essential players in CNS diseases including MS, while macrophage infiltration predominates the CNS inflammatory response in EAE [4].
LPA signals through at least six receptors (LPA 1-6) that exhibit widespread cell and tissue distribution and overlapping specificities [8]. LPARs couple with G-proteins, crucial molecular switches that activate many overlapping signal transduction pathways, leading in pleiotropic effects in almost all cell types, including neuronal ones [8].
ATX expression and extracellular LPA production have been shown to be essential for embryonic development; ATX deficient embryos exhibited severe neural tube defects and impaired neurite outgrowth, that could be restored by LPA ex vivo, indicating a major role for ATX/LPA in CNS development [9]. Moreover, ATX has additional, LPA-independent, roles in CNS development, as it was shown to modulate oligodendrocyte physiology via its matricellular properties and to affect the localization and adhesion of neuronal progenitors [10,11].
In adult healthy life, the CNS is one of the highest ATX expressing tissues, predominately expressing the CNS specific ATX γ isoform which, in comparison to the most abundant β isoform, contains a 25 aa insert of unknown function [12,13]. ATX has been reported to be constitutively expressed by choroid plexus and leptomeningeal cells, releasing ATX in the cerebrospinal fluid (CSF), while increased ATX levels have been detected in activated astrocytes following neurotrauma, as well as in neuroblastomas and glioblastomas [14,15]. PLA2/ATX-dependent LPA/LPAR1 signaling has been shown crucial for the initiation of neuropathic pain [8,16], while deregulated ATX and LPA levels have been detected, in conflicting reports, in the sera and CSF of MS patients [17][18][19][20][21]. Given the established role of ATX/LPA in CNS development, its likely involvement in CNS pathophysiology, as well as the reported LPA effects in different neuronal cell types [8], in this report we examined a possible role of ATX/LPA in the pathogenesis of EAE.

Mice
Mice were housed and bred at 20-22 o C, 55±5% humidity, and a 12-h light-dark cycle at the local animal facilities under specific pathogen-free conditions; water and food were given ad libitum. All reported experimentation in mice for this project, in line with the ARRIVE guidelines, was approved by the Institutional Animal Ethical Committee (IAEC) of BSRC Alexander Fleming and the Veterinary service and Fishery Department of the local governmental prefecture respectively (#449 and #1369/3283/3204 respectively). The generation and genotyping protocols for Enpp2 f/f [9], Enpp2 f/- [9], TgEnpp2 +/+ [22], TgCD11b-Cre [23] genetically modified mice have been described previously. All mice were bred in their respective genetic backgrounds (C57Bl6/J) for over 10 generations. All randomly assigned experimental groups consisted of littermate male age-matched mice. All measures were taken to minimize animal suffering and distress; no invasive or painful techniques were performed requiring anesthetics or analgesics. The health status of the mice was monitored at least once per day; no unexpected deaths were observed. Clinical scoring was reported as indicated in the corresponding figures. Euthanasia was humanly performed in a CO 2 chamber with gradual filling followed by exsanguination, at predetermined timepoints.

Pathology and Immunohistochemistry
Mouse spinal cords were embedded in OCT and cryopreserved at -80 o C. 7 μm sections were sliced transversely into super frost glass slides. Sections were prepared and rehydrated for Luxol fast blue staining of myelin and counterstained with hematoxylin/eosin (H&E) according to standard protocols [26].
Immunocytochemistry sections were left to dry and then were fixed in 4% paraformaldehyde for 20 min at room temperature. Sections were then permeabilized with 0.2% Triton-X for 5 min for intracellular antigen detection wherever it was necessary. Non-specific antigen sites were blocked with blocking solution (Zytomed) for 5 min, followed by addition of rabbit anti-mouse ATX (1:500, Cayman and/or Sigma) or rabbit IgG isotype control antibodies in 2% BSA at 4°C overnight. All washes were performed using PBS-Tween 0.05%. The following day the anti-rabbit Alexa555 (Abcam, 1:1000) secondary antibody was applied to the sections for 1h at room temperature, followed by counter-staining with DAPI (Fluoroshield with DAPI histology mounting medium, Sigma).
Histology images were obtained using a Nikon Eclipse E800 microscope (Nikon Corp., Shinagawa-ku, Japan) attached to a Q Imaging EXI Aqua digital camera, using the Q-Capture Pro software. Immunofluorescence images were captured under a Zeiss Axiovert200 microscope (Carl Zeiss, Oberkochen, Germany).

Western blot
Mouse spinal cords were flushed from the spinal column, snap frozen in liquid nitrogen and stored at -80 o C thereafter. Tissue was homogenised with a glass-glass homogeniser in lysis buffer containing protease inhibitors leupeptin, pepstatin and phenylmethanesulfonylfluoride. Following centrifugation at 17000 g the supernatant (cytoplasmic and soluble proteins) was collected for analysis with western blotting.
Protein concentration was determined with the Bradford assay using a standard curve of BSA (0,125-2mg/ml). Proteins were separated by 8% SDS-PAGE and transferred to Protran nitrocellulose membranes (GE Healthcare, Bucks, UK) using the Trans-Blot SD Semi-Dry Transfer system (Bio-Rad Laboratories, CA, USA). Primary anti-ATX Ab incubation (monoclonal 4F1, 1:1000) was performed overnight in 2.5% (wt/vol) non-fat milk at 4°C. The membranes were then washed three times with TBS-Tween 0.05% and incubated with an anti-rat HRP-conjugated secondary Ab (1:1000) for 1 h at room temperature. Membranes were washed three times with TBS-Tween 0.05%, and antibody-antigen complexes were revealed using luminol as a chemiluminescent reagent.

Flow cytometry
Mononuclear cell suspensions of spinal cords were prepared according to a standard protocol [27]. Briefly, spinal cord tissue was homogenized with 1X HBSS in the

Statistical analysis
In EAE, statistical significance at specific timepoints between experimental groups was assessed with Mann-Whitney Rank Sum Test using SigmaPlot 11.0 (Systat Software, IL, USA). Otherwise, following confirmation of a normal distribution, one-way ANOVA with post hoc correction or student t-test were used accordingly and as indicated in each figure legend. Values are presented as means (±SEM). * indicates a statistical significance difference between the indicated groups (p <0.05).

Increased ATX activity and LPA levels in the plasma upon EAE
To explore a possible involvement of the ATX/LPA axis in the pathophysiology of EAE, C57Bl6/J male mice were immunized with myelin oligodendrocyte glycoprotein (MOG), following a widely used EAE protocol (Fig. 1A) [2]. Immunized mice were weighted (Fig. 1B) and monitored macroscopically daily for clinical signs of EAE for 22 days, in comparison with naïve littermate mice, as shown for the onset (d7), peak (d15) and remission (d22) timepoints ( Fig.1 A, C), where blood plasma and spinal cord tissue samples were collected.
Increased ATX activity levels were detected in the plasma of mice at the peak of EAE clinical symptoms with a well-established enzymatic assay (TOOS; Fig. 1D). To correlate ATX expression levels with its enzymatic substrate and product, we then performed HPLC/MS/MS lipidomic analysis in the plasma. Plasma total LPA levels were also found increased ( Fig. 1E and S1A). No statistically significant deregulation of LPC total plasma levels was recorded ( Fig. 1F and S1A), while no direct correlation of LPC with LPA species was observed (Fig. S1), as is also the case for healthy conditions [5].
To examine if the increased ATX/LPA levels in the circulation are sufficient to modulate EAE pathogenesis, EAE was induced in the heterozygous complete knock out mouse for ATX (Enpp2 +/-), that presents with 50% of normal serum ATX/LPA levels [9], as well in homozygous transgenic mice overexpressing ATX in the liver (TgEnpp2 +/+ ), driven by the human α1-antitrypsin inhibitor (a1t1) promoter, resulting to 200% normal serum ATX/LPA levels [22]. No differences were observed between Enpp2 +/mice (Fig. 2 A-C) or TgEnpp2 +/+ mice (Fig. 2 D-F)

Increased ATX levels and deregulated lipid homeostasis in the inflamed spinal cord
In order to examine if LPA homeostasis and signaling are perturbed upon EAE locally in spinal cords, where EAE predominately manifests in this model [2], we next examined ATX and LPA levels, as well as the mRNA levels of several key players in LPA metabolism and signaling.
Increased ATX (α-ε) mRNA levels were detected with Q-RT-PCR (using two different sets of primers amplifying different parts of the mRNA) in the spinal cords of mice throughout the development of EAE (Fig. 3A), peaking at the onset and remission phases; increased mRNA levels were also detected for the CNS-specific ΑΤΧ γ isoform at the same disease's timepoints (Fig 3A). No direct correlation of ATX mRNA expression profile was apparent with the mRNA expression of inflammatory or fibrotic genes ( Fig. S2 A-C), that could possibly explain the observed expression profile.
A similar ATX expression profile was obtained upon western blot analysis of spinal cord homogenates with a highly specific monoclonal antibody ( Fig. 3B and S3A).
Immunofluorescent staining of ATX in transverse spinal cord cryo-sections confirmed elevated ATX protein levels at the remission phase (but not the onset), and localized staining in inflamed white matter lesions during EAE progression (Fig. 3C); a similar staining profile was obtained with yet another antibody (Fig. S3B). Therefore, EAE development is accompanied by increased ATX levels in the spinal cord, peaking at the remission phase. However, it cannot be excluded that some ATX staining could be due to soluble ATX bound to the cell surface by integrins or other transmembrane or membrane associated molecules [28][29][30], whose expression could be modulated upon EAE.
Increased total LPA levels were also detected with HPLC/MS/MS in the spinal cord of mice at the remission phase of EAE (Fig. 4A), predominated, unlike the plasma, by the 18:2 species (Fig. S1A). The differences in total LPC levels in the spinal cord upon EAE did not reach statistical significance (Fig. 4B), while there was no correlation with the corresponding LPA species (Fig S1A). No significant changes in the mRNA profile of PLPPs, largely responsible for extracellular LPA degradation [31], were noted (Fig. 4C), suggesting minor involvement in the regulation of spinal cord LPA levels in EAE. Furthermore, the mRNA levels for the different receptors of LPA were found to fluctuate during EAE development (Fig. 4D), suggesting exacerbated LPA signaling in the inflamed spinal cord.
Moreover, and given the suggested interplay of the PLA2/LPC and ATX/LPA axes in pathophysiology [32] we next examined the mRNA expression levels of different PLA 2 isoforms. Increased mRNA levels of PLA2 g4a, g4c, g6, g7 and g15 were detected (Fig. S4A), suggesting de novo LPC production in the inflamed spinal cord that could stimulate ATX expression locally; PLA2-mediated LPA production cannot be excluded [7]. The mRNA levels of COX-1/2, guiding the synthesis of proinflammatory eicosanoids from PLA2-synthesized arachidonic acid (AA), were also found increased in EAE (Fig. S4B), indicating the parallel activation of multiple lipid pathways. Accordingly, a deregulation of various lipids was detected in the spinal cords of EAE mice, such as sphingomyelins and ceramides, as well as unsaturated fatty acids

ATX is expressed, among others, from activated macrophages and microglia during EAE pathogenesis
Given the important role of microglia and macrophages in EAE pathogenesis [4] and the suggested pathologic role of ATX expression from macrophages in modeled pulmonary inflammation and fibrosis [33], we next examined if inflammatory macrophages or resident microglia express ATX upon EAE development. Towards that end, FACS analysis of spinal cord homogenates was performed, utilizing the most widely accepted, but not exhaustive, FACS classification set [27]. The results (Fig. 5 A-B) indicate that upon EAE induction both microglial cells (CD45 low CD11b + ), as well as infiltrating blood-derived myeloid cells (CD45 hi CD11b + ) express ATX. Moreover, most macrophages and microglia expressing ATX upon EAE were positive for F4/80 expression (Fig. 5C), a well-known macrophage activation marker [4], suggesting that ATX/LPA could have an autocrine role in macrophage/microglia activation and/or maturation upon EAE.

ATX genetic deletion from CD11b + cells decreases EAE severity
In order to explore a possible pathogenic role of ATX expression from macrophages/microglia during EAE, ATX was genetically deleted in these lineages by mating the conditional knockout mouse for ATX (Enpp2 f/f ) [9] with a transgenic mouse line expressing the Cre recombinase under the control of the CD11b promoter (TgCD11b-Cre) [23]. CD11bEnpp2 -/mice were born with a mendelian ratio and had properly recombined the Enpp2 gene as qualitatively shown with PCR of DNA extracted from spinal cord sections (Fig. 6A).
Genetic deletion of ATX from monocytic cells, decreased mRNA levels in the spinal cord (Fig. 6B), as well as ATX activity levels in the plasma (Fig. 6C), confirming both ATX expression from CD11b + cells, as well as efficient genetic targeting.
CD11bEnpp2 -/mice did not exhibit any behavioral phenotype, nor any appreciable effect in CNS gross morphology under healthy conditions; no major effect in EAE incidence (Fig. 6D) or onset (Fig. 6E) was observed either. However, EAE progression and severity in CD11bEnpp2 -/mice was significantly attenuated, with mice presenting with less disability at the peak of disease (Fig. 6E), resulting in a decreased cumulative score (Fig. 6F); no effect of the conditional Enpp2 allele or the transgenic expression of the cre recombinase alone was observed in littermate controls in multiple experiments ( Fig. 6 D-F). Histological analysis of CD11bEnpp2 -/mice spinal cords indicated decreased inflammation, as detected with H/E staining, decreased demyelination, as detected with luxol staining, as well as significantly less activated astrocytes (GFAP + ), an EAE hallmark (Fig. 6G).
Therefore, ATX expression from CD11b + cells plays an important role in disease severity, likely including the LPA-mediated autocrine activation of macrophages, central to EAE pathogenesis.

Discussion
In this report, increased ΑΤΧ and LPA levels were found in the plasma and spinal cords of mice undergoing EAE development, amid an overall deregulated lipid homeostasis.
More importantly, CD11b + cells, mostly macrophages and microglia, were shown to express ATX upon EAE in the spinal cord, likely resulting to their autocrine activation.
Genetic deletion of ATX from CD11b + cells ameliorated the progression of EAE, thus proving an overall detrimental role of monocytic ATX expression and LPA signaling in EAE pathogenesis.
Increased ATX/LPA levels were detected in the plasma at the peak of EAE ( Fig.1 and S1), declining to normal levels during remission. Noteworthy, even further decreases of LPA were reported for the second remission phase (35 days post MOG immunization) in a SJL EAE model [21]. In MS patients, ATX/LPA plasma measurements have been conflicting [17][18][19][20][21], complicated by treatment history, the timing of sample collection and the available control samples; however, ATX/LPA serum/plasma levels were found higher during relapses comparing to remissions, in the largest and more recent studies [19,21].
Given the suggested LPA effects in endothelial physiology [8,32], the increased BBB permeability upon EAE development [2,3], and the observations made here that ATX/LPA peak in the plasma earlier than in the spinal cord, it is tempting to assume that increased amounts of ATX/LPA could be extravasated in the CNS, possibly contributing to local LPA levels and promoting disease (EAE/MS) pathogenesis.
However, and non-withstanding possible effects due to genetic modifications, no differences were observed between Enpp2 +/or TgEnpp2 +/+ mice and their littermate controls in EAE pathogenesis (Fig. 2), suggesting that a systemic 2-fold (but life-long) fluctuation of ATX/LPA levels per se is likely not a disease modifying factor in the EAE model. Moreover, ATX has been suggested to modulate lymphocyte trafficking in lymphoid organs, where is highly expressed from the endothelial cells of high endothelial venules (HEV) [12,34,35]. However, possible ATX/LPA effects on lymphocyte trafficking, that could complement the well-recognized and therapeutically Increased ATX, mRNA and protein, expression in the spinal cords of mice were detected during the development of EAE (Fig. 3). ATX levels were found significantly higher at the remitting phase of the disease (Fig. 3), rather than its peak as detected in the plasma. A similar mRNA expression profile was also found for the ΑΤΧ γ isoform (Fig 3A), which is predominantly expressed in the CNS [13], suggesting that at least one (undetermined) part of the increased levels of ATX is derived locally from resident CNS cells. Little is known on the regulation of ATX expression in the CNS and especially in the context of EAE. TNF and IL-6, intricately linked with EAE development [24,25], have been reported to stimulate ATX mRNA expression in different, non-CNS, cell types [6,36,37]; both TNF and IL-6 mRNA levels were found increased in EAE (Fig. S2A), suggesting that they could promote ATX expression in some cellular types, in a spatiotemporal manner.
LPC, the substrate of ATX, has been shown to directly promote ATX expression in hepatocytes [37], while it is well known to promote demyelination both in vitro and in vivo [32]. Although LPC is also a signaling lipid, many of the previously reported LPC effects, following the discovery of the lysophospholipase D properties of ATX, are now attributed to ATX and the conversion to LPA. Increased expression of different PLA2 isoforms, largely responsible for LPC synthesis (and possibly some LPA), were detected in the spinal cord upon EAE (Fig. S4A), although no significant changes were detected in spinal cord and plasma LPC levels (Fig. 4B, S1A). However, de novo synthesis could be masked by the high concentrations of LPC. Intriguingly, a search for putative functional partners of ATX at STRING returns mostly PLA2 isoforms, suggesting a physical association of ATX and PLA2s, possibly at the cell surface (via integrins and phosphatidylcholine respectively), and intriguing possibilities in the regulation of local LPC and LPA production and utilization during EAE pathogenesis, especially since integrin-bound-ATX-mediated LPA is thought to be engaged by the adjacent LPA receptors.
Despite the short life of LPA and its likely local consumption, that can complicate conclusions from LPA measurements, the LPA expression profile in the inflamed spinal cord followed the ATX one (Fig. 4), consistent with the consensus notion that ATX is largely responsible for the majority of extracellular LPA production [5][6][7]. No correlation was observed with plasma LPA both in terms of predominating species and as well as of timing, arguing against a major plasma contribution to spinal cord LPA levels; however, it cannot be excluded. A similar LPA spinal cord expression profile has been reported for the SJL EAE model [21], while increased LPA levels in the spinal cord have been also reported upon contusive injury [38], further supporting deregulated LPA homeostasis upon CNS damage.
Among the different potential cellular sources of ATX in the CNS, infiltrating myeloid cells, mainly macrophages, and resident microglial cells (CD11b + cells) were found to express ATX (Fig. 5). In support, lung macrophages post bleomycin-mediated epithelial damage have been shown to produce ATX leading to LPA production in the bronchoalveolar fluid, and promoting the pathogenesis of pulmonary fibrosis [33].
LPS-mediated TLR activation of monocytic THP-1 cells was also reported to lead to ATX production [39]. Moreover, it was recently reported that LPA stimulates the expression of F4/80, a well-known macrophage activation marker [4], in bone marrowderived and splenic monocytic (CD11b + ) cells [40]. Accordingly, most macrophages and microglia (CD11b + cells) expressing ATX upon EAE were positive for F4/80 expression (Fig. 5C), suggesting that the pathogenetic effect of ATX expression from monocytic cells in the context of EAE includes their activation. Accordingly, microglial activation has been shown to mediate de novo LPA production in a model of neuropathic pain [41], while intraspinal injection of LPA was suggested to induce macrophage/microglia activation [38].
Macrophage biology in the CNS remains understudied and controversial [4]. It is generally accepted that macrophages and microglia, that were both shown here to express ATX, can have differential roles in the pathogenesis of CNS disorders [4]. In EAE, myeloid cells are thought to have a greater role, promoting disease onset and progression. By contrast, microglia have been suggested to have a beneficial role, especially in the remitting phase, by aiding in tissue repair and remyelination [4]. The observed ATX expression from macrophages would stimulate their activation and effector functions, central to EAE pathogenesis, whereas ATX expression from microglia could promote wound healing and recovery from the disease. Although the overall relative contribution of macrophages and microglia, and the corresponding ATX expression, to EAE severity remains to be further explored, the observed EAE attenuation upon the genetic attenuation of ATX expression from CD11b + cells (Fig.   6), can likely be attributed to the abolishment of its expression from inflammatory macrophages. However, further studies are required, especially since there are other types of CNS resident macrophages, namely perivascular, meningeal and choroid plexus macrophages [4].
Beyond the constitutive ATX expression from choroid plexus and leptomeningeal cells, and the overall pathogenic ATX expression from activated CD11b + cells upon EAE, other cell types can potentially express ATX upon EAE and further modulate its pathogenesis. ATX has been shown to be expressed from oligodendrocyte precursor cells (OPCs) promoting their differentiation, via LPA and/or its matricellular properties [14,[42][43][44]. ATX expression from OPCs during EAE could promote their maturation and the increased production of oligodendrocytes, thus promoting remyelination, in support of a beneficial role for ATX/LPA in EAE pathogenesis. Moreover, activated astrocytes, that orchestrate CNS tissue repair following injury [45], have been suggested to express ATX upon neurotrama [15]. Multiple LPA effects to astrocyte physiology have been reported, including several of the hallmarks of reactive astrogliosis such as cytoskeletal re-organization, proliferation and axonal outgrowth [8]. Finally, and besides the possible ATX autocrine effects, multiple LPA effects in neuronal cell types have been reported [8], further supporting a multifaceted role for the ATX/LPA axis in EAE/MS pathophysiology. Importantly, the overall effect of ATX activity in EAE pathogenesis depends on the LPA spatiotemporal homeostasis, as well as the relative abundance of the different LPA receptors in the different CNS cell types.
Noteworthy, anti-inflammatory roles have been suggested for the LPA receptor 2 on immune cells [46,47], and Lpar2 null mice were found protected from EAE development [21]. The identification of the specific LPA receptors for each cell type and their spatiotemporal activity will be crucial in understanding the multiple LPA cellular effects in the context of EAE pathogenesis.
The attenuation of EAE severity by the genetic deletion of ATX from CD11b + cells, also considering the other likely sources of ATX expression in the inflamed CNS, suggest possible therapeutic benefits of targeting ATX in EAE/MS. Indeed, pharmacologic potent ATX inhibition was recently reported to attenuate the development of EAE [48], timely with the booming discovery of novel ATX inhibitors [49].

Conclusions
The ATX/LPA axis is increasingly recognized as a major player and drug target in different chronic inflammatory conditions and/or diseases. The present study reveals a novel detrimental role for macrophage (CD11b + ) ATX expression in EAE development, urges further studies on LPA homeostasis in MS/EAE pathogenesis and supports potential therapeutic benefits from targeting ATX.

Acknowledgments
The transgenic TgEnpp2 +/+ mouse was kindly provided by G. Mills. We are thankful to M. Denis for and valuable advice on EAE and I. Barbayianni for assistance in the preparation and submission of the manuscript. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.