CB1 and LPA1 Receptors Relationship in the Mouse Central Nervous System

Neurolipids are a class of bioactive lipids that are produced locally through specific biosynthetic pathways in response to extracellular stimuli. Neurolipids are important endogenous regulators of neural cell proliferation, differentiation, oxidative stress, inflammation and apoptosis. Endocannabinoids (eCBs) and lysophosphatidic acid (LPA) are examples of this type of molecule and are involved in neuroprotection. The present study analyzes a possible relationship of the main receptor subtypes for both neurolipid systems that are present in the central nervous system, the CB1 and LPA1 receptors, by using brain slices from CB1 KO mice and LPA1-null mice. Receptor-mediated G protein activation and glycerophospholipid regulation of potential precursors of their endogenous neurotransmitters were measured by two different in vitro imaging techniques, functional autoradiography and imaging mass spectrometry (IMS), respectively. Possible crosstalk between CB1 and LPA1 receptors was identified in specific areas of the brain, such as the amygdala, where LPA1 receptor activity is upregulated in CB1 KO mice. More evidence of an interaction between both systems was that the CB1-mediated activity was clearly increased in the prefrontal cortex and cerebellum of LPA1-null mice. The eCB system was specifically over-activated in regions where LPA1 has an important signaling role during embryonic development. The modifications on phospholipids (PLs) observed in these genetically modified mice by using the IMS technique indicated the regulation of some of the PL precursors of both LPA and eCBs in specific brain areas. For example, phosphatidylcholine (PC) (36:1) was detected as a potential LPA precursor, and phosphatidylethanolamine (PE) (40:6) and PE (p18:0/22:6) as potential eCB precursors. The absence of the main cerebral receptors for LPA or eCB systems is able to induce modulation on the other at the levels of both signaling and synthesis of endogenous neurotransmitters, indicating adaptive responses between both systems during prenatal and/or postnatal development.


INTRODUCTION
During past decades, lipid molecules present in the CNS were considered components of cell membranes and organelles, intermediary metabolites, or constituents of the energy reservoirs. Currently, it is known that neurolipids are bioactive molecules that can act as neurotransmitters with neuromodulatory properties. Studies analyzing lipid molecules related to signaling systems are clarifying the complex integration of different lipid-metabolic pathways with other physiological functions mediated by lipids. The involvement of lipids in physiological and pathological processes can now also be analyzed by the development of new and potent analytical tools and instruments that enable rapid, high-throughput lipidomic analyses of rodent and human brain tissue sections at anatomical resolution, such as IMS Martinez-Gardeazabal et al., 2017).
Lipid mediators, synthesized from glycerophospholipids, are a class of bioactive lipids that are produced locally through specific biosynthetic pathways in response to extracellular stimuli. These lipid molecules, which hereafter we will call neurolipids, mediate neurotransmission as important endogenous regulators of multiple cerebral processes related to neuroprotective actions, such as neural cell proliferation, differentiation, response to oxidative stress, inflammation and apoptosis. Neurolipids are transported to extracellular compartments (by non-welldefined mechanisms), usually bind to specific GPCRs to trigger intracellular signaling and responses in target cells, and then are sequestered rapidly through enzymatic or non-enzymatic processes (Murakami, 2011). Representative members of this class of neurolipids are eCBs and the LP, LPA.
The first CB receptor cDNA was cloned in 1990 as the first GPCR for lipid mediators (Matsuda et al., 1990). Two CB subtypes have been cloned: CB 1 , the most abundant CB receptor subtype in the CNS, and CB 2 , which is more restricted to glial cells. However, the CB 2 receptor appears to be expressed by some neurons, particularly under certain pathological conditions (Van Sickle et al., 2005;Viscomi et al., 2009;Garcia-Gutierrez and Manzanares, 2011;Navarrete et al., 2012). Moreover, there is evidence for other possible G protein-coupled CBs in the brain (Jarai et al., 1999;Di Marzo et al., 2000;Breivogel et al., 2001). Concerning the other above-mentioned LPs, the first high-affinity cognate cell surface receptor for LPA that was identified as the most abundant in the CNS was LPA 1 , but two additional receptors have been identified (LPA 2 and LPA 3 ). More recently, another three LPA receptors were identified, which were somewhat divergent receptors (LPA 4 , LPA 5 , LPA 6 ) (reviewed in Choi et al., 2010;Yung et al., 2014). These two families of GPCRs, the CB and the LPA GPCRs, are phylogenetically Abbreviations: 2-AG, 2-araquidonoil glycerol; AEA, anandamide; cb WM, cerebellum white matter; CB, cannabinoid receptor; cc, corpus callosum; Cx layer VI, frontal cortex layer VI; eCB, endocannabinoid; GlcCer, glucosylceramide; GPCR, G protein-coupled receptor; ic, internal capsule; LP, lysophospholipid; LPA, lysophosphatidic acid; MALDI-IMS, matrix-assisted laser desorption/ionization imaging mass spectrometry; PA, phosphatidic acid; PC, phosphatidylcholine; PE, phosphatidylethanolamine; ROI, region of interest; TIC, total ion current; VZ, ventricular zone. related. CBs share 18-20% identity of the amino acid sequence with LPA receptors, in particular with LPA 1 , LPA 2 , and LPA 3 (Chun et al., 1999).
Both LPA and eCBs are synthesized mainly by the cleavage of phospholipids (PLs), probably at the cell membrane. The distribution pattern of CB 1 and CB 2 receptors in the developing and adult brain is highly correlated with the expression of the DAGLα/β enzymes, responsible for 2-AG synthesis, suggesting that PL precursors would also be closely localized (Bisogno et al., 2003;Yoshida et al., 2006;Oudin et al., 2011). However, little is known about the specific lipid species that constitute the eCB precursors and particularly, about their specific localization in the brain (Lu and Mackie, 2016). Therefore, one of the aims of the present study is to identify these potential precursors through MALDI-IMS assays. This technique has enriched lipidomic studies due to the new dimensions added by IMS, i.e., the anatomical mapping of the molecules in tissue sections, a key piece of information to understand the physiological roles played by lipids present in the tissue (Caprioli et al., 1997;Jackson et al., 2005;Skraskova et al., 2015;Mohammadi et al., 2016).
Supporting the relationship between the eCB and LPA systems, it is relevant that the endogenous ligands for these systems are closely related, e.g., 2-AG can be metabolized to 2-arachidonoyl-LPA through the action of a monoacylglycerol kinase (Kanoh et al., 1986;Shim et al., 1989). The opposite direction of this phosphorylation reaction, i.e., dephosphorylation, also seems to be possible (Hiroyama and Takenawa, 1998;Nakane et al., 2002). Thus, it appears that 2-arachidonoyl-LPA and 2-AG could be mutually interconverted within a cell, and the direction of the reaction may even differ among the subcellular compartments within the cells and could be modified depending on the availability of substrates, such as ATP (Zhao and Abood, 2013). However, 2-arachidonoyl LPA would not directly interact with the cannabinoid CB 1 receptor, and 2-AG, in turn, would not display affinity for the LPA 1 receptor (Nakane et al., 2002). Moreover, structural analyses of the crystal structure of antagonist-bound LPA 1 receptor predict a functional crosstalk with CB 1 receptors (Chrencik et al., 2015).
During the last decade, the roles of the eCB and LPA systems in the regulation of neuronal progenitor proliferation and differentiation have been described, involving both systems in neurogenesis at the subventricular zone (Birgbauer and Chun, 2006;Galve-Roperh et al., 2006). However, LPA, through binding to the LPA 1 receptor, acts only during the prenatal development of the brain (Hecht et al., 1996). In the meantime, eCBs, through binding to CB 1 and CB 2 receptors, regulate the migration of subventricular zone-derived neuroblasts in the postnatal brain (Oudin et al., 2011). Nevertheless, the eCB system may also control embryonic neuronal development and maturation (Diaz-Alonso et al., 2012).
In addition, both systems display relevant roles in the limbic brain areas and pathways that control emotional processes, such as the amygdala connections, where both systems are involved in the extinction of fear-conditioned responses (Marsicano et al., 2002;Tan et al., 2010;Pedraza et al., 2014). Furthermore, both systems seem to be involved in the control of cognitive processes linked to the hippocampus, such as spatial memory. maLPA 1 -null mice showed, through the LPA 1 receptor-signaling pathway, long-term spatial memory impairment, but not shortterm memory impairment (Castilla-Ortega et al., 2010).
On the other hand, direct hippocampal CB 1 activation by high doses of exogenous or endogenous cannabinoid compounds induce characteristic long-term memory impairment (Puighermanal et al., 2009). In the same way, the genetic deletion of CB 1 receptor induces neuronal loss in CA1 and CA3 areas of the hippocampus, which triggers a decline in cognitive functions (Bilkei-Gorzo et al., 2005).
In summary, both neurolipid systems are implicated in cortical development in the prenatal brain, in the extinction of fear-conditioned behavior in the amygdala, and in the modulation of long-term memory in the hippocampus, suggesting a convergence in the control of similar physiological functions in the CNS. In the present study we analyze the adaptive changes that are probably produced during mouse prenatal and postnatal development in the eCB and LPA neurolipid systems using both maLPA 1 -null and CB 1 KO mice, to analyze the interactions and crosstalk between the CB 1 and LPA 1 mediated-activities and to identify the anatomical localization of the possible precursors of the endogenous ligands for both signaling systems. Technically, the study combines functional [ 35 S]GTPγS autoradiography with MALDI-IMS to analyze consecutive brain slices.

Chemicals
[ 35 S]GTPγS (initial specific activity 1250 Ci/mmol) was purchased from Perkin Elmer (Boston, MA, United States). WIN55212-2 was purchased from Tocris (Bristol, United Kingdom). Oleoyl-L-α-lysophosphatidic acid sodium salt (LPA), 2-mercaptobenzothiazole (MBT), DL-dithiothreitol (DTT), guanosine-5 -diphosphate (GDP), guanosine-5 -o-3triphosphate and β-radiation sensitive Kodak Biomax MR films were acquired from Sigma-Aldrich (St. Louis, MO, United States). The [ 14 C]-microscales used as standards in the autoradiographic experiments were purchased from Amersham Biosciences (Buckinghamshire, United Kingdom) and American Radiolabeled Chemicals (ARC, St. Louis, MO, United States). Finally, for the preparation of the different buffers, the treatment of slides, re-crystallization of the matrix and film development, several additional compounds supplied from different laboratories were used. All the compounds were of the highest commercially available quality for the necessity of the neurochemical studies.

Animals and Tissue Preparation
maLPA 1 -null homozygous (n = 8) and wild-type 3-month-old males (n = 8) (on a mixed background C57Bl/6 × 129SW) were obtained from Hospital Carlos Haya, Málaga, Spain, as a result of a collaboration with Dr. Estivill-Torrús's research group. The maLPA 1 -null (from Málaga variant of LPA 1 -null) mouse colony arose spontaneously from the initially reported LPA 1null mouse line (Contos et al., 2000) while crossing heterozygous foundational parents within their original mixed background since phenotypic variations have been described in almost complete perinatal viability and show a reduced VZ, altered neuronal markers, and increased cortical cell death that results in a loss of cortical layer cellularity in adults (Estivill-Torrus et al., 2008;Matas-Rico et al., 2008).
CB 1 -knockout mice are a strain maintained at the University of the Basque Country and were provided by C. Ledent of the University of Brussels (Belgium). For this study, 9 week-old male CB 1 -knockout (n = 5) and wild-type mice (n = 5) were used. The generation of mice lacking CB 1 CBs was described previously (Ledent et al., 1999). To homogenize the genetic background of the mice, the first generation of heterozygous mice was bred for 30 generations on a CD1 background, with selection for the mutant CB 1 gene at each generation. Mice were housed in a temperature (22 • C) and humidity-controlled (65%) room with a 12:12-h light/dark cycle (light between 08:00 and 20:00 h), with food and water ad libitum. All procedures were performed in accordance with European animal research laws ( Mouse brains were quickly removed under anesthesia. Then, tissues were frozen on dry ice and kept at −80 • C. The brains were cut on a Microm HM550 cryostat (Thermo, Germany) to obtain 20 µm sections that were mounted onto gelatin-coated slides, and these were stored at −20 • C until use.

[ 35 S]GTPγS Binding Assay
The tissue sections were air-dried for 15 min then washed in HEPES-based buffer (50 mM HEPES, 100 mM NaCl, 3 mM MgCl 2 , and 0.2 mM EGTA, pH 7.4) for 30 min at 30 • C in a water bath. The preincubation was repeated in new buffer. In a second step, the slides were incubated for 2 h at 30 • C supplemented with 2 mM GDP, 1 mM DTT, adenosine deaminase (3 µ/l) and 0.04 nM [ 35 S]GTPγS. The agonist-stimulated binding was measured under the same conditions in the presence of the specific GPCR agonists oleoyl-L-α-lysophosphatidic acid sodium salt; LPA (10 −5 M) and WIN55212-2 (10 −5 M). Non-specific binding was determined in the presence of 10 µM non-labeled GTPγS. Sections were washed twice in an ice-cold (4 • C) HEPES buffer (50 mM, pH 7.4), dipped in distilled water, and air-dried. Sections were exposed to autoradiography film (Kodak Biomax MR) together with 14 C standards for 48 h at 4 • C in hermetically closed cassettes.

Quantitative Image Analysis of Autoradiograms
Films were scanned and quantified by transforming the optical densities into nCi/g of tissue equivalent (nCi/g t.e.) using ImageJ software (NIH-IMAGE, Bethesda, MA, United States) (developed at the U.S. National Institutes of Health and freely available on https://imagej.nih.gov/ij/). The [ 14 C] radioactive standards that were co-exposed with the tissue slides were used to calibrate the optical densities with the level of radioactivity labeled to the sections. Experimental data were analyzed by using the computer programs GraphPad Prism (v. 5.0, GraphPad) and Microsoft Excel. Data are expressed as the mean values ± SEM. The basal binding was calculated as nCi/g tissue equivalent, using the values provided in the 14 C microscales. The agonist stimulation of [ 35 S]GTPγS binding is expressed by the percentage of stimulation over the basal (%) in the presence of the different agonists, calculated as: (([ 35 S]GTPγS agonist-stimulated binding) × 100/([ 35 S]GTPγS basal binding)) − 100. The regions of interest (ROI) were defined from the scanned film in the case of the GTPγS autoradiography and directly from the image obtained from MALDI-IMS, using counter stained consecutive sections and a mouse brain atlas, without knowing the genotype of the mice and without taking into account the intensities of the rest of the animals. Differences between regions for each mouse genotype (wild type vs. transgenic) were analyzed by the unpaired two-tailed Student's t test.

Sample Preparation for MALDI-IMS
The original lipid composition and anatomical characteristics of the tissue must be well-preserved throughout the samplepreparation process (Schwartz et al., 2003). The fresh-frozen brains of the mice were cut on a Microm HM550 cryostat (Thermo, Germany) to obtain 20 µm sections and stored at −20 • C until the moment of use.
Once the initial tissue preparation steps were completed, the matrix was deposited on the tissue surface prior to analysis by a sublimation process using the protocol described by Gonzalez de San Roman et al. (2017). For tissue sections mounted on glass slides, sublimation was performed using 300 mg of MBT, which was used as a chemical matrix, to control the deposition time and temperature (30 min, 140 • C). Therefore, it was possible to control the thickness of the matrix layer and optimize the S/N ratio of the mass spectra, avoiding lipid migration on the tissue slice thanks to the lack of solvents using this sublimation procedure.

Mass Spectrometer
A MALDI LTQ-XL-Orbitrap (Thermo Fisher, San Jose, CA) equipped with a nitrogen laser (λ = 337 nm, rep. rate = 60 Hz, spot size = 80 × 120 µm) was used for mass analysis. Thermo's ImageQuest and Xcaliber software were used for MALDI-IMS data acquisition. The images were acquired in both negative and positive ion modes. The positive ion range was 500-1000 Da, with 10 laser shots per pixel at a laser fluence of 15 µJ. The negative ion range was 400-1100 Da, with 10 laser shots per pixel at a laser fluence of 15 µJ. The target plate stepping distance was set to 150 µm for both the x-and y-axes by the MSI image acquisition software. The mass resolution was approximately 100,000 laser shots in both positive and negative ion modes. The data were normalized using the TIC values to avoid potential displacement in the masses along the tissue that may be induced by experimental factors, including irregularities of the surface.

Image and Spectral Analysis for MALDI-IMS
The MALDI-IMS technique was applied to localize and semiquantitatively analyze the different lipid species present in WT and maLPA 1 -null mice in positive and negative ion detection modes. We employed optimized experimental conditions for lipid detection previously set up by our research group and others, including selection of the MBT chemical matrix and deposition mode by sublimation procedures (Astigarraga et al., 2008;Yang and Caprioli, 2011). First, IMS of sagittal mouse brain sections was performed using the positive ion detection mode, and approximately 300 mass peaks were detected in the mass range from 500 to 900 m/z. Then, the same slices of tissue were analyzed in negative ion detection mode using a mass range from 400 to 1100 m/z. A complete mass spectrum was obtained from each pixel along all the surface of each brain tissue slice, obtaining the slice images as a composition of the intensity color of each peak (molecule or m/z) at each pixel. Specifically, an ion of interest can be extracted from the spectra, and an image of that particular ion distribution in the tissue can be visualized. Once we had analyzed at least five sagittal brain slices samples from five different animals of WT and another five samples of maLPA 1null mice, we selected different ROIs, such as hippocampus, cortex, cerebellum, cc, and striatum. The spectrum normalization was calculated using the TIC for each mass to exclude a number of noise spectra. Next, we compared WT and maLPA 1 -null using Origin R 8 software.
In summary, the MALDI images were generated using ImageQuest software (Thermo Scientific). Each of the m/z values was plotted for signal intensity per pixel (mass spectrum) across a given area (tissue section). The m/z range of interest was normalized using the ratio of the TIC for each mass spectrum. The intensity reached by each mass spectrum (intensity of each peak, m/z value or molecule) was further calculated as a ratio of the peak with the highest intensity, and the average spectrum was calculated with OriginPro 8 software (Northampton, MA, United States). The most intense peak was considered 100%. Then, the two-tailed unpaired t test was performed for the comparison of two groups. The results were considered significant if p ≤ 0.05.

Peak Assignment
Analysis of the lipid composition at CNS tissue samples from mice included in the present study was complex. A large number of different possible lipid species were detected, and some of them shared similar masses and were not detected and/or assigned to a specific lipid molecule by previous studies. Nevertheless, the assignment of lipid species was facilitated by the use of the Lipid MAPS database 1 and different previous reports (Berry et al., 2011;Martinez-Gardeazabal et al., 2017), using a 5 ppm mass accuracy as the tolerance window for the assignment. The numbers (x:y) following the glycerolipid species denote the total length and the number of double bonds of the acyl chains, respectively, while the GlcCer species numbers correspond to the length and number of double bonds of the acyl chain added to those of the attached sphing-4-enine (d18:1) or sphinganine (d18:0) base.

RESULTS
[ 35 S]GTPγS Binding Assay of maLPA 1 -Null Mouse Brain Sections The [ 35 S]GTPγS binding stimulated by WIN55212-2 was measured in brain slices of maLPA 1 -null and WT mice to localize and quantify the activity of CB 1 CBs. The aim was to observe, in adult mice was the possible modulation of cannabinoid signaling that was induced by the lack of the LPA receptor subtype, LPA 1 , during the animal's previous development. The basal activity (in the absence of agonist) of the G i/o -coupled GPCR in maLPA 1null mice was already shown in Gonzalez de San Roman et al. (2015). The G i/o -coupled CB 1 receptor activity induced by the CB 1 agonist WIN55212-2 was increased at the prefrontal cortex layers I-III (WT 212 ± 19% vs. maLPA 1 -null 369 ± 49%, p ≤ 0.05) and in layer VI (WT 284 ± 14% vs. maLPA 1 -null 396 ± 22%, p ≤ 0.05). The increase in the CB 1 activity was also significant in the cerebellum gray matter (cb GM) of maLPA 1null mice compared to WT (WT 23 ± 15% vs. maLPA 1 -null 107 ± 13%, p ≤ 0.05) and cb WM (WT 678 ± 149% vs. maLPA 1null 1377 ± 207%, p ≤ 0.05). The globus pallidus, a nucleus with a high density of CB 1 receptors in mice, also showed an increase in the activity of these receptors in maLPA 1 -null mice (WT 1705 ± 151% vs. maLPA 1 -null 2037 ± 82%), but this change was not statistically significant (Table 1 and Figure 1). The [ 35 S]GTPγS binding stimulated by oleoyl-L-α-lysophosphatidic acid sodium salt (LPA) showed no stimulations in any brain area of the maLPA 1 -null mice, demonstrating the specificity of the experimental binding procedure and the right phenotype of the animals (Gonzalez de San Roman et al., 2015).

[ 35 S]GTPγS Binding Assay in CB 1 KO Mouse Brain Sections
We compared the [ 35 S]GTPγS binding stimulated by LPA in CB 1 KO and WT mice to observe the consequences of the developmental absence of the main eCB receptor on the signaling through cerebral GPCRs for the LPA 1 receptors. There were statistically significant changes that were restricted to the amygdala and showed increased [ 35 S]GTPγS binding induced by LPA (WT −20 ± 4.3% vs. CB 1 KO 32 ± 6.3%, p ≤ 0.05). The LPA 1 receptor activity of CB 1 KO mice was comparable to WT mice in all the other brain areas that were analyzed (Table 2 and Figure 2). The [ 35 S]GTPγS binding stimulated by WIN55212-2 showed no stimulation in any brain area of the CB 1 KO mice, demonstrating the specificity of the experimental binding procedure and the right phenotype of the animals (data not shown).

MALDI-IMS in CB 1 KO Mouse Brain Sections
The results of the MALDI-IMS experiments using CNS slices from WT and CB 1 KO mice were also performed in positive and negative ion detection modes. As has been described, we used an optimized protocol for the lipid detection. The analysis of IMS on sagittal mouse brain slices was performed in positive ion detection mode in a mass range from 500 to 1000 m/z and in negative ion detection mode from 400 to 1100 m/z. We selected the ROI and followed the steps described in the "Materials and Methods" section. The results indicated  that some lipid species were modified in the brain of CB 1 KO mice in both positive and negative ion detection modes. Some lipid species were present at a higher intensity than in WT mouse brain. These lipid species included PC(32:0) + in hippocampus (WT 37.6 ± 3.6% vs. CB 1 KO 65.3 ± 3.1%, p < 0.01) and cortex (WT 33.2 ± 3.7% vs. CB 1 KO 54.7 ± 6.5%, p < 0.05) and PC(34:1) + also in hippocampus (WT 48.6 ± 5.0% vs. CB 1 KO 67.7 ± 3.5%, p < 0.05) and cortex (WT 51.2 ± 4.8% vs. CB 1 KO 69.5 ± 4.5%, p < 0.05). Both PC molecules were increased in the frontal cortex, hippocampus and striatum but were not modified in the cerebellum of the CB 1 KO mice. In contrast, the relative intensities of other lipid species were lower in CB 1 KO mice

DISCUSSION
The present study shows specific anatomical alterations of a few lipid species that are potential PL precursors of LPA or eCBs, analyzing simultaneously in the same animals the regulation of the signaling by their specific CB 1 and LPA 1 receptors in KO mice for each of these receptors, which are the respective predominant subtypes present in the CNS. The analyzed regions were selected attending to previously described brain areas showing high densities of LPA 1 or CB 1 receptors. The analysis of the activity of CB 1 receptors in maLPA 1 -null mice allowed us to detect a regulation mechanism in both in the frontal cortex and in the cerebellum. The absence of LPA 1 in maLPA 1 -null mice results in a reduction of neural cells in layers II/III, V, and VI, increased apoptosis, alterations in the formation of the cerebral cortex and premature expression of neuronal markers (Estivill-Torrus et al., 2008). The eCB system is a critical component of frontal cortex physiology. The CB 1 receptor is mainly expressed in layers II/III and V/VI (Lafourcade et al., 2007), same areas where there is a depletion of neurons in maLPA 1 -null mice. Therefore, the increased functional activity of CB 1 receptors suggests a compensatory effect of the eCB system to balance the neuronal reduction. Furthermore, the cerebellum is another brain area where we detected an increase in CB 1 receptor activity in maLPA 1 -null mice. Cannabinoids modulate the proliferation of cerebellar neuronal precursors via the CB 1 receptor (Trazzi et al., 2010). Hence, the lack of the TABLE 3 | MALDI-IMS intensities (expressed as % of the most abundant peak) of molecular species of PC and PA that were found modified in positive ion detection mode in sagittal brain sections from WT (n = 5) compared to the maLPA 1 -null (n = 5).
LPA 1 receptor could lead to an adaptive regulation in CB 1mediated activity.
On the other hand, when the activity of LPA 1 receptors was analyzed in CB 1 KO mice, the LPA 1 receptor activity was increased only in the amygdala compared with WT mice. The CB 1 receptor plays an important role in processes mediated by the amygdala, such as emotional responses (Azad et al., 2004;Laviolette and Grace, 2006;Roche et al., 2007;Tan et al., 2010). The eCB system in the amygdala has also been implicated in pain modulation (Manning et al., 2003), fear conditioning (Marsicano et al., 2002), anxiety (Imperatore et al., 2015) and fear-conditioned learning . The LPA 1 receptor plays a crucial role in the extinction of fear conditioning (Pedraza et al., 2014) and has been associated with the mixed depressive-anxiety phenotype (Moreno-Fernandez et al., 2017). Our results suggest that the lack of the CB 1 receptor leads to an increased LPA 1 receptor activity in the amygdala, as a compensatory effect exerted by LPA signaling in this limbic area. Considering the present results, it is difficult to hypothesize why no other brain regions have been regulated in CB 1 KO mice, this regional exclusivity to the amygdala emphasizes the already reported role of LPA 1 receptors in fear extinction, that might be indicating a close physiological control in anxiety behavior and fear-conditioned learning together with CB 1 receptors. However, other functions modulated by the eCB system, such as pain or movement control, are possibly not being compensated by LPA 1 signaling.
The lack of LPA 1 receptor expression during mouse development led to an increase in the activity of the CB 1 receptor in the cortical area and cerebellum, and the lack of the CB 1 receptor led to an increase in LPA 1 receptor activity in the amygdala. Therefore, the absence of neurotransmissions mediated by CB 1 or LPA 1 receptors in each strain of knockout mice converged to a common compensatory mechanism between both systems in some specific brain areas (cortex, cerebellum and amygdala). The reported compensation would affect receptors coupled to other types of G proteins, however, as previously mentioned, we were only able to analyze those coupled to the G i/o subtype. It is improbable that CB 1 receptors play a role since previous results reported a similar distribution regarding the CB 1 receptor autoradiographic densities compared to functional autoradiography stimulated by specific agonists such as the WIN55212-2 (Llorente-Ovejero et al., 2018). However, it has not been yet possible to get the distribution of LPA 1 receptors.
Moreover, the co-regulation of eCBs and LPA must be yielded during the pre-and postnatal development of both strains of mice.
The mechanisms underlying the observed modulation of CB 1 receptors in LPA 1 -null mice may involve an increase in the expression of CB 1 receptors mediated by another subtype of LPA receptors coupled to G i/o proteins (see Supplementary Material).
The metabolic precursors of the endogenous agonists for both neurolipid systems, eCB and LPA, are PLs present at the cellular membranes. The metabolism of PLs such as PC, PS and PE is thought to be responsible for the synthesis of LPA and the eCBs AEA and 2-AG. Therefore, we also hypothesized that the identification of specific PL modifications in the brain of CB 1and LPA 1 -KO mice would help us to understand the molecular neuroscience of the interactions between both systems. Thus, the PL precursor localization was an important step to clarify the synthesis of these lipid mediators. The application of the MALDI-IMS technique to maLPA 1 -null and CB 1 KO mice allowed us to determine the anatomical localization of each lipid species in brain tissue slices. Unlike classical neurotransmitters, eCBs and LPA are not stored in specific compartments of the cells, such as vesicles, but instead are produced mainly "on demand" by stimulus-dependent cleavage of membrane PL precursors (Schmid et al., 1990;Di Marzo et al., 1994;Sugiura et al., 1995;Aoki, 2004). However, little is known about which are the specific lipid species for the synthesis of the endogenous agonists LPA, AEA, and 2-AG, or about their anatomical localization within the brain.
The results obtained with the MALDI-IMS assay in maLPA 1null and WT mice showed differences in the intensity of some lipid species, mainly in the positive ion detection mode. The affected species were PA(16:0/18:1) + K + and four different PCs, including PC(32:0), PC(36:1) + K + , PC(38:6) + K + , and PC(38:4) + K + , whose more probable acyl chains were assigned based on previous studies and were identified as 16:0, 18:0, 18:1, 20:4, and 22:6 (Sugiura et al., 2009;Berry et al., 2011;Martinez-Gardeazabal et al., 2017). Although much has been learned about the physiological roles of LPA through the analyses of LPA receptors, the mechanism by which LPA is produced remains unclear. At least two pathways have been postulated. In the first pathway, LPA would be produced from PL by the autotaxin enzyme. In the second pathway, PA would be generated from PL or diacylglycerol and then deacylated by different phospholipases, FIGURE 4 | MALDI-IMS images representing the different lipid species that showed statistically significant differences in the detection intensities between WT and CB 1 KO mice. The spectra were recorded with 150 µm spacing. Ten shots at each point were accumulated. Laser fluency was set at 15 µJ. such as PLA 1 or PLA 2 (Aoki, 2004;Aoki et al., 2008). The predominant fatty acyl moieties identified for LPA in rat brain are 18:1 (46.9%), 18:0 (22.5%), and 16:0 (18.8%), although a relatively small, but significant, amount of a 20:4-containing species has also been detected (7.2%) (Das and Hajra, 1989;Sugiura et al., 1999). PC species with arachidonic acid as its acyl chain PC(38:4), which was significantly increased in maLPA 1 -null mice in the cerebellum, could be a precursor of LPA (20:4). This species could be converted to 2-AG through LPA-phosphatase and trigger the increase in CB 1 receptor activity (Nakane et al., 2002). Moreover, PC(36:1) is significantly increased in most of the analyzed areas and tended to increase in the maLPA 1 -null mice in brain areas where LPA 1 receptors are specifically distributed, suggesting that PC(36:1) is a potential precursor for LPA synthesis (Gonzalez de San Roman et al., 2015;Mihara et al., 2019).
Regarding the MALDI-IMS results in CB 1 KO and corresponding WT mice, the ratio of some lipid species was changed compared to WT mice, in both positive and negative ion detection modes. The affected species were PCs, PEs and PSs. PC(32:0) and PC(34:1) were increased approximately 50% in CB 1 KO mice in some areas, such as the hippocampus and frontal cortex. We assumed that changes in PC(32:0) and PC(34:1) indicated that the lack of the CB 1 receptor led to an increase in the abundance of lipids with palmitic acid. Both PCs can contain palmitic acid at fatty acyl moieties, and the CB 1 receptor is palmitoylated for efficient coupling to a specific subset of G proteins located on membrane lipid rafts with a high concentration of saturated fatty acids (Oddi et al., 2012).
The synthesis of AEA and 2-AG could be produced by the cleavage or hydrolysis of membrane PL precursors. For the synthesis of 2-AG, at least three pathways have been proposed (Higgs and Glomset, 1994;Nakane et al., 2002;Kano et al., 2009). The synthesis of AEA seems to involve the enzymatic hydrolysis of the corresponding N-acylphosphatidylethanolamines (NAPE) (reviewed in Schmid et al., 1990Schmid et al., , 1996Hansen et al., 1998). In view of our results, we infer that the observed decrease in PE species in CB 1 KO compared with WT mice could be related to precursors of AEA, due to the requirement of PE molecules for AEA synthesis. This decrease may be a modulation of PEs due to the lack of the target, the CB 1 receptor. In addition, the decrease in PS species could be derived from the PE diminution because PE is one of the sources for PS synthesis (Vance and Tasseva, 2013). However, no changes in AEA or 2-AG levels have been reported between CB 1 KO and WT mice (Maccarrone et al., 2001). The undergoing neuro-adaptive metabolic changes could lead to an increase in the activity of the enzymes responsible for the production of AEA from PE in KO mice, but obtaining the same levels of AEA than in WT.
Globally, the absence of LPA 1 or CB 1 receptors is able to induce modulation on the other system at the levels of both signaling and synthesis of endogenous neurotransmitters, suggesting adaptive responses between both systems during prenatal and/or postnatal development. Future goals would be to further clarify the relationship between the cannabinoid and lysophosphatidic systems as neurolipid systems with potential neuroprotective actions to identify new therapeutic treatments that could be developed for neurodegenerative diseases, such as Alzheimer's disease.