Enteric Neuropathy Can Be Induced by High Fat Diet In Vivo and Palmitic Acid Exposure In Vitro

Objective Obese and/or diabetic patients have elevated levels of free fatty acids and increased susceptibility to gastrointestinal symptoms. Since the enteric nervous system is pivotal in regulating gastrointestinal functions alterations or neuropathy in the enteric neurons are suspected to occur in these conditions. Lipid induced intestinal changes, in particular on enteric neurons, were investigated in vitro and in vivo using primary cell culture and a high fat diet (HFD) mouse model. Design Mice were fed normal or HFD for 6 months. Intestines were analyzed for neuronal numbers, remodeling and lipid accumulation. Co-cultures of myenteric neurons, glia and muscle cells from rat small intestine, were treated with palmitic acid (PA) (0 – 10−3 M) and / or oleic acid (OA) (0 – 10−3 M), with or without modulators of intracellular lipid metabolism. Analyses were by immunocyto- and histochemistry. Results HFD caused substantial loss of myenteric neurons, leaving submucous neurons unaffected, and intramuscular lipid accumulation in ileum and colon. PA exposure in vitro resulted in neuronal shrinkage, chromatin condensation and a significant and concentration-dependent decrease in neuronal survival; OA exposure was neuroprotective. Carnitine palmitoyltransferase 1 inhibition, L-carnitine- or alpha lipoic acid supplementation all counteracted PA-induced neuronal loss. PA or OA alone both caused a significant and concentration-dependent loss of muscle cells in vitro. Simultaneous exposure of PA and OA promoted survival of muscle cells and increased intramuscular lipid droplet accumulation. PA exposure transformed glia from a stellate to a rounded phenotype but had no effect on their survival. Conclusions HFD and PA exposure are detrimental to myenteric neurons. Present results indicate excessive palmitoylcarnitine formation and exhausted L-carnitine stores leading to energy depletion, attenuated acetylcholine synthesis and oxidative stress to be main mechanisms behind PA-induced neuronal loss.High PA exposure is suggested to be a factor in causing diabetic neuropathy and gastrointestinal dysregulation.


Introduction
Obese and/or diabetic patients face an increased risk of multiple complications and decreased quality of life. [1] In a population based survey more than 75% of diabetic patients report gastrointestinal (GI) symptoms. [2] The basis for optimal GI regulation is the enteric nervous system (ENS) innervating the entire digestive tract, and pivotal in coordinating motility, secretion and blood flow.
A patients metabolic and nutritional status contribute to the development of type 2 diabetes (T2D) and obesity is considered the single largest risk factor. [1] The lipid profile in obese and type 2diabetic patients is characterized by elevated plasma levels of the saturated free fatty acid (FFA) palmitic acid (PA, 16:0) and the monounsaturated FFA oleic acid (OA,18:1). [3] Lipotoxicity is induced by prolonged exposures to an excess of FFA, particularly PA. Its hallmarks are lipid accumulation in nonadipose tissues, cellular dysfunction and apoptosis. [4,5] Lipid-induced ENS neuropathy and subsequent pathophysiological signs of GI dysfunction are scarcely studied.
Several animal models exist that display various aspects of T2D. However, except for a few studies, showing reduced concentrations of regulatory peptides, [6] loss of cholinergic and nitrergic neurons, and altered intestinal transit, [7,8] there have been few reports about ENS plasticity in metabolic disorders. Notable is that diabetics (BMI.30, HbA1c.7%) display loss of enteric neurons, higher incidence of GI symptoms, and reduced glutathione levels indicating oxidative stress as the key event. [9] These reports support the need for further investigations of obesity and T2D related effects on ENS, and consequent intestinal dysfunction.
The aims of the present study were to investigate long-term effects of high fat diet (HFD) on intestinal remodeling and survival of enteric neurons. Additionally, cellular survival and intracellular signaling pathways after exposure to PA and/or OA in mixed myenteric cultures were studied.

Ethics statement
Procedures were approved by the animal ethics committee, Lund and Malmö, SE. Animals were used in accordance with the European Community Council Directive (2010/63/EU) and the Swedish Animal Welfare Act (SFS 1988:534).

Animals and tissue preparations
Male and female littermate C57BL/6 mice (n = 21), from in house breeding facilities were used in feeding experiments. Mice aged 1 month were divided into two groups; the normal diet (ND) group continuing on standard diet (Lactamin R36)(n = 8) and the HFD group changing to a diet containing 45% kcal from fat (New Brunswick Research diets D12451, USA) (n = 13) (table 1). After 6 months mice were sacrificed by cervical dislocation, weighted and body composition determined by dual-energy x-ray absorpmetry (PIXImus, GE lunar, USA). [10] The GI tract from cardia to rectum was collected, opened along the mesenteric border, emptied and weighed. Segments of ileum and transverse colon were fixed and processed for cryo sectioning. [11] Female Sprauge-Dawley rats (n = 46, 170-260 g), (Charles River, DE) were used for in vitro experimentation. Primary cell cultures of the longitudinal smooth muscle layer with adherent myenteric ganglia from small intestine were prepared as described previously. [12] The resulting mixed cell cultures were first grown 4 days in medium (neurobasal A, containing 10% fetal bovine serum, 0.5 mM L-glutamine, 50 U/mL penicillin and 50 mg/mL streptomycin, all from Life Technologies, SE). Fresh medium containing applicable experimental test agents was then added and incubation for an additional period of 4 days followed. Control wells were cultured in parallel. Cells were fixed and subjected to immunocytochemistry. [12] Pharmacological agents and experimental in vitro set-ups Stock solutions of PA, methyl palmitic acid (MPA) and pharmacological agents (table 2) were prepared, aliquoted and stored at 220uC.
PA and MPA were conjugated to BSA (fraction V, Merck, SE) by adding PA-or MPA stock together with 20% BSA w/v to medium, and mixed at 37uC, 1 h prior to use. Final PA:BSA and MPA:BSA molar ratios were 4-5:1. Final BSA concentration was #2% v/v. OA was purchased pre-conjugated to 2% BSA. Different concentrations (10 24 -10 23 M) of albuminated FA were added to cultures, controls received medium containing 2% BSA.

Immunocytochemistry and histochemistry
For details on primary and secondary antibodies see table 3. All antibodies were diluted in phosphate buffered saline containing 0.25% Triton X-100 and 0.25% BSA (PBS-T-B). For visualization and quantification of submucous and myenteric neurons, cryosections from ND and HFD mice were immunolabeled with biotinconjugated HuC/HuD antibodies and vectastain ABC kit (Vector laboratories Inc., USA), according to manufacturer's protocol. Morphometric analyses of small and large intestine were on toludine blue (0.1% in 60% ethanol) stained cryosections. Intracellular lipid accumulations were visualized using BodipyH 493/503 (Life Technologies, SE) diluted 1:1000 in PBS-T-B, incubated 1 h, washed in PBS-T and mounted in ProLongHgold (Life Technologies, SE).
Double immunolabeling of cultures were by overnight incubation in moist chamber at 4uC with a mixture of primary antibodies. Secondary antibodies were mixed and incubated 1 h at RT. Hoechst (Life Technologies, SE) cell nuclei counter staining was performed according to manufacturer's protocol. Intracellularly accumulated lipid droplets were visualized with BodipyH 493/503 added to secondary antibodies at final dilution of 1:1000. Mounting was in PBS:glycerol 1:1 followed by fluorescence microscopy (Olympus BX43, LRI, SE) with appropriate filter setting.

EdU incorporation
Proliferating cells were identified by Click-iT EdU Alexa flour 488 imaging kit (Life Technologies, SE), according to manufacturer's protocol. EdU was added to culture wells 24 h prior to fixation.

Morphometic analyses and cell calculations
Small and large intestinal mucosa and muscularis propia thicknesses were estimated for each mouse using the mean from ten representative measurements (Aperio ScanScope CS/GL SS5082 and ImageScope). The numbers of HuC/HuD-immunoreactive submucous and myenteric neuronal cell bodies were counted per mm section length. From each mouse, cryosections (10 mm) cut longitudinally at 3-4 different depths comprising a total length of 25-30 mm, were used.
Neuronal survival in vitro, after exposure to the various treatments, was calculated by counting the total number of surviving neurons in the entire culture well (69 mm 2 ) and expressed as percentage of the number of neurons in the control well run in parallel. Survival of glia, muscle cells and proliferating EdU positive cells were, due to their abundance, calculated by capturing (Olympus cellSens) five randomized microscopic fields (center and each quadrant, 0.45 mm 2 per field) from every culture well (n = 3-38). From these selections the total cell number, using Hoechst staining, and the total number of either S100 or alpha smooth muscle actin (aSMA) immunoreactive (expressed as percentage of control) or EdU positive cells (expressed as percentage of total) were estimated. Calculations were done using Image J cell calculator (rsb.info.nih.gov/ij/). Measurements on soma and nuclei area, cellular perimeter as well as on neuronal chromatin condensation were done using the Olympus cellSens system. Twenty cells of each type were randomly identified by immunocytochemistry in each culture well and analyzed, n = 5 in each group.

Statistical analyses
Data are presented as means 6 SEM and analyzed by GraphPad Prism (GraphPad Software Inc, USA). Statistical significances were determined using two-way-analysis-of-variance, followed by Bonferronis post hoc test (in vivo data), or one-wayanalysis-of-variance followed by Dunnet's post hoc test towards controls (in vitro data). A confidence level of 95% was considered significant.

Data availability
We agree to make freely available any materials and data described in the publication that may be reasonably requested for the purpose of academic, non-commercial research.

HFD compared to ND
Mice fed HFD displayed increased body weight (p,0.001) and body fat percentage (p,0.001) compared to ND mice. Weights of the GI tract did not differ between the groups (ND 3.3 g60.3, HFD 3.6 g60.3). In mice fed HFD, compared to littermates on a ND, there were significant reductions in the numbers of myenteric neuronal cell bodies in both ileum and colon (ileum p,0.01, colon p,0.001), on the other hand the numbers of both ileal and colonic submucous nerve cell bodies were unchanged. In ileum of HFD mice, the mucosal thickness was reduced compared to ND mice (p,0.01), while no difference was noted in the muscularis propria. Colon mucosa and muscularis propria thicknesses were identical in  Table 3. Overview of primary and secondary antibodies used in immunocytochemistry.

Survival of myenteric neurons, smooth muscle cells and glia in vitro
Primary cultures of the outer muscle layer with adherent myenteric ganglia from rat small intestine include neurons, glia and intestinal smooth muscle cells. All experimentations comprised 4 days recovery culture period, followed by 4 days of treatment. Controls were always run in parallel and cellular survival estimated as percentage of control.
Effects of PA, OA and MPA on cellular survival. Survival after exposure to PA (10 24 -10 23 M), OA (10 24 -10 23 M) or PA (4610 24 M)+OA (10 23 M) were determined. Control wells displayed a large number (4.360.2/mm 2 , n = 43) of evenly distributed neuronal cell bodies with rich terminal networks. Addition of PA severely reduced, while OA increased, the number of surviving neurons in a concentration-dependent manner. Addition of PA+OA resulted in neuronal survival similar to control. Numerous glia was present in control (355630.5/mm 2 , n = 24) and no changes were observed after PA and/or OA treatments. In control wells muscle cells were frequent (795672/ mm 2 , n = 22) and provided a coalescent layer. After PA or OA treatment muscle cells disappeared in a concentration-dependent manner. Notable was that the combined treatment of PA+OA resulted in an increased survival of muscle cells compared to control. Figure 2 and table 4 illustrate and summarize findings on PA and/or OA treatment.
MPA is a nonmetabolizable form of PA. Neuronal survival in cultures treated with MPA (10 24 -10 23 M) did not differ from controls (table 5). All wells displayed evenly distributed nerve cell bodies, with terminal networks, without signs of morphological alterations. Also muscle cells and glia were unaffected in appearance and numbers after MPA treatment (table 5).    Peroxisome proliferator-activated receptor (PPAR) activation. OA-induced neuroprotection in central neurons involves activation of PPAR alpha. [16] It was tested if this was true for enteric neurons as well. Presence of PPAR antagonist GW6471 (10 27 -3610 26 M) did not reverse OA-induced neuroprotection or loss of muscle cells. GW6471 alone did not change cellular survival nor did it affect OA-evoked rescue of PAinduced neuronal and muscle cell losses.
Ceramide formation. Inhibiting PA from entering the ceramide pathway by adding serine palmitoyltransferase inhibitor myriocin (10 212 -10 210 M) did not reverse PA-induced neuronal or muscle cell losses. Myriocin alone reduced survival of muscle cells but did not change that of neurons or glia.
Palmitoylcarnitine formation. The carnitine palmitoyltransferase 1 (CPT1) inhibitor etomoxir (10 27 -10 25 M) prevents conversion of PA-CoA to palmitoylcarnitine in mitochondrial, peroxisome and endoplasmatic reticulum (ER) membranes. Its presence reversed PA-induced neuronal, but not muscle cell losses in a concentration-dependent manner. Etomoxir alone did not change survival of any of the cultured cell types.
L-carnitine (10 24 -10 22 M) supplementation to cultures concentration-dependently reversed PA-induced neuronal loss. Lcarnitine alone was neuroprotective (p,0.01), but had no effect on glia or muscle cell survival, nor did it reverse PA-induced loss of muscle cells.
Reactive oxygen species (ROS) and ER stress. Catalase is the main enzyme handling hydrogen peroxide. Presence of catalase (100-3000 U/mL), did not reverse PA-induced neuronal or muscle cell losses. Catalase alone did not change cellular survival. Besides hydrogen peroxide ROS includes peoxynitrite, hydroxyl, peroxyl and hypochlorus radicals as well as singlet oxygen. These were scavenged using the antioxidant, ALA (10 24 -10 23 M) which concentration-dependently reversed PA-induced neuronal, but not muscle cell losses. ALA alone increased neuronal survival (p,0.05), reduced muscle survival but did not change survival of glia. To test possible contributions of nitric oxide or peroxynitrite on PA-induced neuronal loss N-nitro-L-arginine methyl ester (L-NAME), a general inhibitor of nitric oxide synthetases, was used. L-NAME (10 23 M) alone did not change neuronal survival nor did it rescue neurons from PA-induced loss.
Elevated levels of ROS and oxidative stress may cause ER stress. The chaperone 4-phenyl-butyric acid (4PBA) is a stabilizor of protein folding. 4PBA (5610 24 -10 22 M) did not reverse PAinduced neuronal or muscle cell losses. 4PBA alone did not change cellular survival.
Excitotoxicity. PA metabolism may lead to excess formation of glutamate, causing neurotoxicity by hyperactivation of NMDA receptors. Ketamine (3610 25 -10 23 M), a NMDA receptor blocker, did not reverse PA-induced neuronal or muscle cell losses. Ketamine alone did not change cellular survival.

Discussion
Long-term intake of HFD caused reduced numbers of myenteric neurons in ileum and colon, a thinner ileal mucosa and lipid accumulation in muscularis propria. The mechanisms by which prolonged exposure to lipids and lipid metabolites cause neuropathy and morphologic remodeling were addressed in the current study.
The dominating FA's in HFD, PA and OA, were tested in vitro for their ability to mimic the neuropathy and lipid handling noted in vivo. While PA evoked a detrimental effect on neurons OA was clearly neuroprotective. Both PA and OA caused intestinal muscle cell loss, but unchanged glial survival. In addition PA caused shrinkage and chromatin condensation of neurons and transition of glia to a rounded phenotype. Substituting PA for nonmetabolizable MPA did not affect cellular survival. This strongly suggests PA-induced neuronal and muscle cell losses are due to uptake and intracellular handling of PA.

Mechanisms behind PA-induced neuronal loss
PA-induced neurodegeneration occurs in neuronal cell lines and is suggested to be caused by oxidative and ER stress. [17,18] In the present study the underlying mechanisms for PA-induced myenteric neuronal loss were elucidated by intervening pharmacologically with several intracellular pathways in lipid metabolism. The results indicate that ceramide and hydrogen peroxide formation as well as glutamate generated excitotoxicity all could be excluded as underlying mechanisms. The studies pointed towards energy depletion, halted acetylcholine synthesis and oxidative stress as major events. The CPT1 inhibitor etomoxir, L-carnitine and the important enzymatic co-factor and antioxidant ALA, [19] all protected myenteric neurons from PA-induced loss. Although attenuation of PA stimulated beta-oxidation would be expected to result in reduced ROS production, this pathway is less likely as neurons, including ENS, are not considered relying on beta-oxidation for energy production. A condition reflected in low neuronal expression levels of CPT1a and b. [20,21] On the other hand CPT1c is abundantly expressed and localized to the ER, [21,22] suggesting biosynthetic rather than energy related functions. CPT1c mediates PA-CoA entry into ER lumen where it catalyzes the formation of palmitoylcarnitine from PA-CoA. [22] Palmitoylcarnitine is amphipathic and has pleiotropic effects including membrane destabilization, lysis, reduced membrane potential, calcium release and caspase activation. [23,24] CPT1 relies on Lcarnitine as a co-carrier of acyl-CoA and an increased PA load would drain cellular L-carnitine stores resulting in decreased availability of acyl-CoA and thus in a decreased formation of acetylcholine. [25,26] Limiting the ability of neurons to synthesize acetylcholine would compromise maintenance and trigger neuronal loss. Another important aspect of exhausted neuronal Lcarnitine stores is impediment of glucose oxidation resulting in decreased energy production. Depleting the mitochondrial acetylcarnitine levels decreases pyruvate dehydrogenase activity and increases acetyl-CoA levels resulting in attenuation of the Krebs cycle. [26,27] Reversal of PA-induced neuronal loss by ALA can be explained by its vital role, as a co-enzyme, in the pyruvate-and alpha-ketoglutarate dehydrogenase complexes. [28] In addition to its metabolic role ALA is also a powerful antioxidant, [19] and part of it rescue potential on PA-induced neuronal loss is most likely by ROS scavenging. ROS are important cellular redox regulators, fostering the unfolded protein response (UPR) and ER stress, eventually leading to programmed cell death. [29,30] In the present study PA-induced neuronal loss was unaltered by presence of the chaperone 4PBA, thus excluding UPR-mediated ER stress. Oxidative stress may still be a culprit, as palmitoylcarnitine and altered phospholipid formation increase ER permeability and ROS discharge. [31,32] The protective mechanism of etomoxir on PA-induced neuronal loss is likely explained by its CPT1 inhibition. PA-CoA accumulation is, in contrast to palmitoylcarnitine, not amphipathic, and  forms phospholipids and/or other lipid intermediates. [33] Neuronal rescue by exogenously added L-carnitine can be explained by replenishment of available L-carnitine stores.
AMPK activation by the AMP analog AICAR caused myenteric neuronal loss in vitro, comparable to that of PA. Simultaneous addition of AICAR and PA did not further augment the neuronal loss, leading to an initial assumption that PA, as shown in a hypothalamic cell line, [34] activated AMPK. This seems unlikely though as ALA and etomoxir, both able to counteract PA-induced neuronal loss, were unable to reverse AICAR-or AICAR plus PA-induced neuronal loss. From this we conclude that AMPK-induced neuronal loss apparently occurs in myenteric neurons. This pathway is, however, not activated by PA and not executed through activation of CPT1.
Damaged neurons display various morphological signs of death. [35,36] PA exposure enhanced chromatin condensation within cultured myenteric neurons, suggesting programmed cell death. Approximately 18% of all cultured neurons displayed activated caspase-3 immunoreactivity irrespective of treatment (own unpublished observations). Since this finding is not in accord with the observed neuronal loss we question if activated caspase-3 can be considered an optimal marker for apoptosis in myenteric neurons. This conclusion is supported by other studies on peripheral neurons. [11,13,37] Mechanisms behind OA-induced neuroprotection OA is, by way of PPAR-alpha activation and upregulated expression of GAP43, reported to protect cultured neurons from PA-induced loss. [16] In the present work we identified a marked neuroprotective capacity of OA per se on myenteric neurons; this effect was not attenuated by the presence of PPAR inhibitor. In addition OA significantly counteracted both PA-and AICARinduced neuronal loss. OA mediated reversal of PA-induced neuronal loss did not involve PPAR activation, while its reversal of AICAR-induced neuronal loss did. This further supports that PAand AICAR-induced neuronal loss are differently executed. OA's reversal of PA-induced neuronal loss is suggested to involve redirection of PA towards undamaging lipid intermediates, as suggested in other tissues. [38][39][40] PA and OA effects on intestinal smooth muscle cells Striated and vascular muscle cells efficiently take up fatty acids in vitro. [39,41] PA exposure results in apoptosis, [41,42] ER stress and abnormal lipid distribution that can be counteracted by OA. [39,41] The beneficial effects of OA have been ascribed to its promotion of triglyceride and lipid droplet formation and decreased formation of phopholipids, resulting in low lipotoxicity. [38,39] Conflicting data suggests OA to be both apoptotic and anti-apoptotic. [40,43,44] Current results showed that OA alone, like PA, reduced survival of intestinal muscle cells. It is important to note that simultaneous treatment with PA and OA failed to execute the apoptotic and anti-proliferative responses obtained with each FA alone. Ceramide-and ROS-formation, ER stress and beta-oxidation as well as AMPK-and PPAR-activation were excluded as causing PA-or OA-induced muscle cell loss. We suggest that the combined PA and OA treatment promoted maintenance of intestinal muscle cells by dual mechanisms; a sustained proliferation and shuttling or redistribution of lipids towards formation of triglycerides. The latter suggestion is based on the finding that simultaneous presence of PA and OA dramatically increased lipid droplet formation in muscle cells. Interestingly, this is well in line with the here reported observation that HFD mice, in contrast to ND mice, harbored numerous lipid droplets in muscularis propria.

PA effects on glia
Survival of glia in vitro was unchanged by PA, OA, ALA, and AICAR treatments. PA notably altered glia morphology. The cytoplasmatic expansion of glia may be in response to PA-induced muscle cell loss, ensuring a supportive layer for the neurons, or rely on the neuronal loss. Astrocytes, closely resembling enteric glia, take on a flat and extended shape when cultured alone, but adapt a stellate shape when co-cultured with neurons. [45] A rounded glia phenotype is suggested to accompany alterations in chloride homeostasis, a process coupled to physiological requirements. [45] Present data does not allow for speculations on altered chloride currents or the consequences thereof, but suggests phenotypic changes to be secondary to PA-induced loss of neurons and/or muscle cells.

Conclusions
Long term intake of HFD caused a remarkable loss of intestinal myenteric neurons. HFD contains substantial amounts of PA and OA; putative candidates for mediating the HFD-induced neuropathy. As tested in vitro PA was capable of mimicking HFD-induced neuronal loss while OA was neuroprotective. PA or OA-exposure both caused intestinal muscle cell loss while glia numbers were unchanged. PA-induced neuronal loss is suggested to be multifaceted. The present data suggests intracellular metabolic derangements due to L-carnitine depletion and palmitoylcarnitine formation to be central. This ultimately results in lack of energy substrates, reduced acetylcholine synthesis, membrane deterioration and oxidative stress. The mechanism leading to PA-or OAinduced loss of intestinal muscle cells is suggested to be unbalanced lipid metabolism, as simultaneous exposure to PA and OA left intestinal muscle cells unaffected in number and with a high intracellular lipid droplet accumulation. The latter probably serving as a reservoir halting lipid metabolism. An increment in intramuscular lipid accumulation, as noted in HFD mice, may explain their well-preserved muscularis propria.  This is the first study showing that long term intake of HFD leads to ENS neuropathy and an imminent GI dysmotility in rodents. Based on the here presented in vitro studies we suggest PA as a putative candidate mediating this. Whether these findings can be extrapolated into the clinic needs further investigations.