Myriocin and D-PDMP ameliorate atherosclerosis in ApoE-/-mice via reducing lipid uptake and vascular inflammation.

Sphingolipids have been implicated in the etiology of atherosclerosis. The commonly used sphingolipid inhibitors, myriocin (a ceramide inhibitor) and D-PDMP (a glycosphingolipid inhibitor), have shown therapeutic potential but their efficacy and their underlying mechanisms remain unclear. Here, apoE-/- mice were fed a high-fat diet and treated with a control, myriocin, D-PDMP, or atorvastatin for 12 weeks. We analyzed the effects of these drugs on the size and detailed composition of atherosclerotic plaques. Molecular biological approaches were used to explore how the inhibitors affect lipid metabolism and foam-cell formation. Treatment with myriocin or D-PDMP led to smaller and less vulnerable atherosclerotic lesions and was almost as effective as atorvastatin. Sphingolipid inhibitors downregulated the expression of MCP-1 and its receptor CCR2, which play a key role in monocyte recruitment. They also decreased pro-inflammatory Ly-6chighmonocytes and influenced the uptake of modified LDL by downregulating the expression of CD36 and LOX-1. The inhibitors exhibited the advantage of maintaining normal glucose homeostasis compared with atorvastatin. These findings reveal for the first time that the modulation of sphingolipid synthesis can effectively alleviate atherosclerosis progression by preventing lipid uptake and reducing inflammatory responses in the arterial walls.

using a glucometer (Sinomedisite, Beijing, China). The area under the curve (AUC) was calculated using the 116 trapezoidal method.

118
Additional blood samples (20 μL) for the determination of insulin levels were collected at baseline. An Ultra-

119
Sensitive Mouse Insulin ELISA kit (Crystal Chem, Downers Grove, IL, USA) was used, following the 120 manufacturer's instructions. Whole-body insulin resistance (IR) was estimated using the homeostatic model

147
Qualitative and quantitative analyses were performed using a Q Exactive Orbitrap mass spectrometer (Thermo 148 Scientific, CA, USA). A full-scan spectrum was recorded over the 240-2000 m/z range, and the qualitative 149 assignment of sphingolipids was carried out in the MS/MS mode. The resulting data were analysed using 150 Tracefinder (Thermo Scientific). The mass tolerances of precursor and fragment ions were 5 and 10 ppm, 151 respectively.

154
Following euthanasia, mice were perfused with 10 mL ice-cold normal saline. The livers were dissected and 155 weighed. The middle lobes of livers were embedded in optimum cutting temperature compound (OCT) (Sakura,

Assessment of oxidative stress in liver
166 Liver tissue samples (approx. 100 mg) were sonicated and homogenized in normal saline on ice. Oxidative stress 167 markers were measure using a total antioxidant capacity (T-AOC) assay kit, a micro reduced glutathione (GSH) 168 assay kit, a superoxide dismutase (SOD) assay kit (WST-1 method), a catalase (CAT) assay kit (Ultraviolet), and 169 a malondialdehyde (MDA) assay kit (TBA method). All kits came from Nanjing Jiancheng Bioengineering 170 Institute, China.

173
The aortic arches were carefully isolated under a stereo microscope (Leica, Wetzlar, Germany) and 174 photographed. The entire aortas were then dissected and fixed in 4% paraformaldehyde. Before being stained 175 with Oil Red O for en face analysis, periadventitial fat and connective tissue were carefully removed using 176 microforceps and microscissors. For the photography, the whole aorta was opened longitudinally and pinned to a 177 black cystosepiment surface.

179
For the aortic sinus analysis, aortic roots were dehydrated and embedded in paraffin. Starting from the 180 appearance of the aortic valve, sequential sections 4-μm thick were collected across the aortic root (approx. 350 181 μm in total). For each mouse, sections of 100-μm intervals were used to evaluate the size of atherosclerotic 182 lesions. The severity of plaques was classified as: early lesions, with early fatty streaks; moderate lesions, with a 183 collagenous cap; and advanced lesions, with an increased necrotic area and involvement of the media.

184
Morphological analysis of collagen contents in the lesions was performed by staining with Movat Pentachrome.

185
The lipid area was defined as Masson's trichrome stain-free. Images were captured under a Nikon Eclipse Ti2 186 microscope (Kyoto, Japan). Plaque necrotic core areas were manually drawn. Quantification was completed 187 using IPP software. where the three aortic valve leaflets appeared simultaneously. The sections were incubated with primary 192 antibodies, followed by horseradish peroxidase (HRP)-conjugated secondary antibodies (Supplemental Table   193 1), and then developed using DAB substrate (brown).

196
Briefly, the sections of aortic root were blocked in PBS with the appropriate serum (10%) for 1 h, incubated at 4 ℃ 197 overnight with primary antibodies, and then stained with fluorescent-dye-conjugated secondary antibodies for 1 198 h at 37 ℃ (Supplemental Table 1). Mounting medium with DAPI (ZSGB-Bio, Beijing, China) was used to 199 stain the nuclei. Images were captured and processed using identical settings in a Nikon A1MP inverted confocal  Table 2. PCR amplification consisted of an initial denaturation step at 95 °C for 10 min, followed by 40 cycles 208 of PCR at 94 °C for 5 s and 60 °C for 40 s (CEX Connect Real-Time System, Bio-Rad, Hercules, CA, USA).

209
Quantification of relative mRNA expression was calculated using the efficiency-corrected 2 -△△CT method. Actin 210 gene expression was used as the internal control, and the differences were presented as fold-changes relative to 211 the vehicle group.

222
After blocking with 5% non-fat dried milk, the membranes were incubated with primary antibodies overnight at 223 4 ℃, and then with appropriate secondary antibodies at room temperature for 1 h. Blots were detected using a 224 chemiluminescence detection system (G:BOX Chemi XT4, Syngene) and viewed in Image J software for data 225 analysis. All protein levels were normalized to β-actin or α-tubulin signals. Uncropped scans of western blots are 226 provided in Supplementary Figures 2 and 3.

229
Statistical analyses were performed using SPSS v.22.0 software (IBM Corp., Armonk, N.Y., USA). All 230 continuous variables are presented as the mean ± SEM and statistical significance was accepted when P < 0.05.

231
One-way ANOVA followed by a least significant difference (LSD) post hoc test (equal variances assumed) or 232 Dunnett's post hoc test (no equal variances assumed) was applied for comparisons among the groups.

233
Pearson's chi-squared test was used to verify the correlation between certain sphingolipids and the intervention    Figure 1A). Myriocin and D-PDMP were well-tolerated and caused no overt side effects.

243
One animal in the atorvastatin group exhibited hair loss, low body-mass and abnormal splenomegaly. At the end 244 of week 24, we performed mouse trans-thoracic echocardiography. Heart function was measured and the ejection 245 fraction and fractional shortening were found to be comparable among the four groups (Supplemental Table 3,

254
To quantify the degree of lumen stenosis, aortic-root sections were stained with H&E, Figure 1F). We found  Figure 4G). Notably, myriocin seemed to show better efficacy than 260 atorvastatin.

262
Next, a more detailed analysis of aortic-root plaque composition was performed. We found that the three groups  Figure 1I). No significant differences were seen 266 by staining for α-SMA, a biomarker of smooth muscle cells (SMCs) ( Figure 1I). Fewer lipid areas were 267 observed in the plaques but their collagen content was increased by treatment with myriocin, D-PDMP and 268 atorvastatin ( Figure 1J). The necrotic core areas, which are features of vulnerable plaques, showed significant 269 reduction ( Figure 1K). On the basis of these results, it was shown that mice receiving drug treatment displayed 270 more early-to-moderate lesions, with fewer severe lesions ( Figure 1L).

273
Following myriocin, D-PDMP and atorvastatin treatment, we evaluated the ability of these drugs to lower the  (Figures 2A, B). This result was consistent 279 with an earlier finding by Garner et al. that treatment with another glycosphingolipid inhibitor, EtDO-P4, did not 280 elicit significant changes in plasma TC or TG levels(9). However, our results were inconsistent with those of 281 Chatterjee and colleagues, who found that treatment with D-PDMP significantly decreased LDL-C (~70%) and 282 TG (~50%) and increased HDL-C (~3 times)(10). There are several possible explanations which could account 283 for these differences. First, both our group and Garner et al. fed mice a high-fat diet (0.15% cholesterol),

284
whereas Chatterjee et al. fed mice a western-style diet (1.25% cholesterol). The TG levels (~790 mg/dl) in 285 vehicle group from Chatterjee's data were obviously higher than the levels in this present study (~100 mg/dl) 286 and Garner's study (~140 mg/dl). Second, Chatterjee and colleagues treated their apoE -/mice for 24 weeks, 287 twice the duration of drug treatment in our study. In addition, the lower TG in our study may be partially related 288 to fasting state, as prolonged fasting reduced TG levels.

290
The effects of the various treatments on the serum sphingolipidome were determined by Orbitrap-based high  Table 4A, 4B). In this study, we also found that D-PDMP impeded

311
To investigate how the different treatments affected lipid metabolism, we first tested their liver histopathology.

312
As expected, none of the interventions significantly affected the level of hepatocyte damage-markers (AST and 313 ALT) in peripheral blood (Figures 3A and B). However, the mass of the livers in myriocin-treated mice . Although it has been reported that the use of genetic engineering approaches or specific inhibitor treatments 318 to decrease sphingolipids could reverse hepatic steatosis in obese rodents(14, 15), these findings were not 319 observed in our study. Atorvastatin treatment appears to be the most effective for attenuating hepatic lipid 320 accumulation in apoE -/mice.

322
Next, we performed an analysis of mRNAs involved in cholesterol metabolism (Hmgr, Ldlr, Pcsk9, Srebp2 and 323 Acat1; Figure 4A), fatty-acid uptake (Cd36, Fabp1 and Caveolin-1; Figure 4B (Figures 4B, E). Western blot analyses confirmed the reductions in CD36, Caveolin-1 and SCD-1 330 proteins (Figures 4G, H and J). Although some studies have suggested that myriocin can reduce serum TG and 331 that D-PDMP plays a role in promoting cholesterol efflux and triglyceride metabolism(6, 10), we found no 332 evidence to support these claims in the present study (Figures 4C, D and I). The promotion of fatty-acid 333 oxidation was seen in atorvastatin-treated mice ( Figure 4F).

335
Taken together, these findings suggest that sphingolipid inhibitors do not significantly affect cholesterol 336 metabolism and that their protective role in fatty-acid metabolism is not sufficient to reverse liver lipid-337 deposition. These results also hinted that the main mechanism linking sphingolipid inhibitors to decreased 338 atherosclerosis in apoE -/mice is probably not involved in the hepatic clearance of excessive lipids but may 339 instead be involved in vascular lipid uptake or other pathways.  (ACAT1; Figures 5G, H and M). In addition, the main lipid droplet-coating protein in 371 foam cells, perilipin2, was detected. Perilipin2 has been shown to play an important role in the formation of 372 atheroma(18). In the present study, the downregulation of perilipin2 by sphingolipid inhibitors at the mRNA 373 level was significant compared with vehicle-treated animals, but at the protein level there were no statistically 374 significant differences among the groups (Figures 5N and O).  (7), but other reports have suggested that myriocin does not appear to 381 ameliorate the activation of hepatic inflammation(19). Since little is known about the role of myriocin or D-PDMP in vascular inflammation or apoptosis, we hypothesized that these two inhibitors could improve plaque 383 stability by attenuating the inflammatory signaling cascade.

385
The circulating concentration of high-sensitivity C-reactive protein (hs-CRP), an inflammatory biomarker, was 386 significantly decreased by both D-PDMP and atorvastatin (Figures 6A). Similar reduction was observed in 387 aortic protein MAC-3, a marker for macrophages (Figures 6B). Myriocin also led to a downregulated trend, with 388 less hs-CRP and MAC-3. To obtain additional lines of evidence, we next analyzed the expression of genes 389 involved in vascular inflammation. The mRNA levels of pro-inflammatory cytokines (Il-1β, Il-6 and Tnf-α), 390 chemokines (Ccl-2), and adhesion molecules (Icam-1 and Vcam-1) were in general downregulated by the two 391 inhibitors ( Figure 6C). Immunofluorescence staining for IL-1β in atherosclerotic lesions and immunoblotting 392 for the VCAM-1 protein showed a corresponding downregulation (Figures 6D and E). The marker of 393 angiogenesis, vascular endothelial growth factor (VEGF), showed a similar change at the mRNA level but no 394 significant difference was found at the protein level ( Figure 6E). Besides, D-PDMP induced the up-regulation of 395 Il-10, an anti-inflammatory cytokine ( Figure 6C). Collectively, myriocin treatment reduced aortic inflammatory 396 responses, while D-PDMP treatment exerted a robust anti-inflammation function, although it could not correct 397 the abnormally high serum cholesterols in apoE -/mice.

399
Since myriocin and D-PDMP showed anti-inflammatory properties, we also tested their anti-oxidative properties,  (Figures 6F and G). We next investigated aortic mRNA expression of antioxidant 403 enzymes, such as superoxide dismutase (SOD) and catalase (CAT), and the subunits of NADPH oxidase 404 responsible for superoxide formation, such as p22 phox , gp91 phox , p47 phox , p40 phox and p67 phox (Figures 6H and I).
405 Liver tissues were also tested to further validate the anti-oxidative roles. We found that D-PDMP and 406 atorvastatin reduced hepatic oxidative stress, as evidenced by increased levels of T-AOC and glutathione (GSH) 407 and decreased levels of malondialdehyde (MDA) (Figures 3K, L and M). However, the hepatic levels of SOD 408 and CAT were not significantly elevated (Figures 3N and O). Taking these findings together, D-PDMP was 409 able to reduce oxidative stress impairment, but this function was not clearly observed in myriocin-treated mice.  (21). The activation of apoptotic pathways was investigated but neither 420 cleaved PARP or caspase-12 expression was significantly changed (Figures 6N and O).

Myriocin and D-PDMP were beneficial for maintaining whole-body glucose homeostasis
423 No significant differences in fasting blood-glucose levels were observed among the four groups (Figure 7A), or 424 in the oral glucose tolerance test (OGTT) at the end of week 24 (Figure 7B and C). However, glucose levels in 425 the mice were elevated along with the HFD feeding. Notably, the fasting insulin level and the insulin resistance 426 (IR) index in myriocin-treated mice and D-PDMP-treated mice were lower compared with atorvastatin-treated 427 mice (Figures 7D and E). The serum level of adiponectin, an anti-diabetic adipokine, was relatively elevated,

428
indicating an improvement in systemic insulin sensitivity ( Figure 7F). Additionally, taking into account the fact 429 that the phosphorylation of AKT plays a key role in insulin signaling, immunoblots of phosphorylated and total

430
AKT from the liver were performed. Compared with the vehicle-control group, p-AKT levels were elevated in 431 the livers of animals treated with myriocin or D-PDMP ( Figure 7G). Taking these findings together, and 432 considering the association between statins and diabetes, these new sphingolipid inhibitors may probably be an 433 adjunct to or a replacement for statin therapy.

436
We observed similar curves for body mass and comparable body masses among all treatment groups 437 (Supplemental Figure 5A). Mice treated with myriocin ate 6.5% less food during the first month compared 438 Downloaded from https://portlandpress.com/clinsci/article-pdf/doi/10.1042/CS20191028/868337/cs-2019-1028.pdf by guest on 27 February 2020 with the vehicle-treated animals (Supplemental Figure 5B). Nevertheless, there were no significantly different 439 food efficiencies among the groups (Supplemental Figure 5C). A previous study reported that myriocin 440 decreased the total fat mass in obese mice(22). Herein, we evaluated the mass of white adipose tissue (WAT),

441
including gonadal, perirenal and subcutaneous WAT (gWAT, pWAT and sWAT) and brown adipose tissue 442 (BAT) in apoE -/mice. We also performed histological analyses. The cross-sectional area of gWAT in myriocin-443 treated mice was noticeably smaller (Figure 7H). Myriocin decreased the mass of WAT after normalizing for 444 body mass (Figures 7I, L and M). However, we observed a trend of increasing BAT mass in these mice 445 ( Figure 7N). F4/80 immunohistochemistry staining confirmed that fat pads in mice from the myriocin-treated 446 group had far fewer infiltrated macrophages compared with those of vehicle-treated mice, indicating a reduction 447 in adipose inflammation (Figures 7J and K).

449
This study is aimed to elucidate the therapeutic efficacy of sphingolipid inhibitors in improving atherosclerosis.

466
Smaller plaque sizes represent reduced lipid accumulation, which is associated with monocyte recruitment and 467 foam-cell formation. We were intrigued to find out whether these sphingolipid inhibitors might be involved in 468 skewing monocyte differentiation, since the Ly-6c high phenotype has been reported to express higher levels of 469 CD36 and possess a greater capacity to form foam cells(16). We discovered for the first time that

487
Therefore, myriocin and D-PDMP were shown to be capable of improving atherosclerosis as a result of their 488 anti-inflammatory properties.

490
Numerous data support the pronounced efficacy of "sphingolipid reduction therapies" for improving insulin 491 resistance in obese mice (13,26). We observed a slight improvement in glucose metabolism in mice treated with 492 myriocin and D-PDMP compared with statin-treated mice. It was previously reported that obese mice treated 493 with myriocin displayed decreased adiposity, which was associated with increased energy expenditure(22). We 494 found similar adipose remodeling in myriocin-treated apoE -/mice: their gWAT mass was substantially reduced 495 and their BAT mass was relatively increased. To the best of our knowledge, iminosugar N-(5'-adamantane-1'-yl-496 Downloaded from https://portlandpress.com/clinsci/article-pdf/doi/10.1042/CS20191028/868337/cs-2019-1028.pdf by guest on 27 February 2020 methoxy)-pentyl-1-deoxynojirimycin (AMP-DNM) has been reported to improve adipose function (27). In this 497 study, reduced numbers of large adipocytes and a reduced inflammatory state were observed in D-PDMP-treated 498 adipose tissue, but these findings were not statistically significant compared with the control group. A recent 499 study reported that statins cause cell-autonomous impairment in insulin-stimulated adipocyte lipogenesis(28).

500
Here, the mass of atorvastatin-treated fat tissue displayed a similar downward trend, but the number of F4/80-501 positive cells did not noticeably decrease. The underlying regulatory mechanisms remain incompletely 502 understood.

504
There were a number of limitations to this study. We cannot rule out the possible contribution of apolipoprotein 505 E-deficiency to whole-body metabolism. In addition, the possible side effects of the drugs, such as muscle pain 506 due to statin therapy, were not adequately taken into account. Statin treatment in combination with myriocin or 507 D-PDMP will be further analyzed in future studies.

509
Overall, myriocin and D-PDMP treatment were sufficiently effective at reducing and stabilizing atherosclerotic 510 plaques. Unlike atorvastatin, their athero-protective roles were not primarily dependent on the reduction of 511 serum cholesterols. This mechanistic study indicated that sphingolipid inhibitors reduce the formation of foam 512 cells as a result of decreased lipid uptake, rather than enhanced reverse cholesterol transport. Furthermore, their 513 anti-inflammatory roles were also proved. Myriocin and D-PDMP appear to be an effective replacement for or 514 adjunct to statin therapy, which would possibly provide a two-pronged approach for treating atherosclerosis.

515
Moreover, these sphingolipid inhibitors performed better at maintaining a normal glucose level compared with 516 atorvastatin and therefore we postulate that, in clinical practice, they may be promising avenues to explore for 517 concurrently combating atherosclerosis and diabetes.

531
These sphingolipid inhibitors performed better at maintaining a normal glucose level compared with 532 atorvastatin. Noticeably, we discovered for the first time that pharmacologically inhibiting sphingolipid 533 synthesis plays a role in suppressing monocytes that differentiate toward a Ly-6c high phenotype. Moreover, 534 the modulation of sphingolipid synthesis can effectively alleviate atherosclerosis progression by preventing 535 lipid uptake and reducing inflammatory responses in the arterial walls.

558
The authors declare that there are no competing interests associated with with the manuscript.