Vascular lipidomics analysis reveales increased levels of phosphocholine and lysophosphocholine in atherosclerotic mice

Objective Atherosclerosis (AS) is the major cause of cardiovascular disease, and dyslipidemia is a principal determinant of the initiation and progression of AS. Numerous works have analyzed the lipid signature of blood, but scarce information on the lipidome of vascular tissue is available. This study investigated the lipid profile in the aorta of ApoE−/− mice. Method ApoE−/− mice were randomly divided into two groups: (1) the normal diet (ND) group and (2) the high-fat diet (HFD) group. After feeding for 8 weeks, the plasma low-density lipoprotein (LDL), total cholesterol (TC), and triglyceride (TGs) levels were measured. UHPLC-Q Exactive plus MS was used to assess the lipid profile using both positive and negative ionization modes. Results LDL and TC levels were significantly increased in HFD mice, and lipid deposition, plaque area and collagen fiber levels were increased in HFD group. In addition, a total of 131 differential lipids were characterized, including 57 lipids with levels that were increased in the HFD group and 74 with levels that were decreased. Further analysis revealed that the levels of several differentially expressed phosphocholines (PCs) and lysophosphocholines (LPCs) were significantly increased. These PCs included PC (38:3), PC (36:4), PC (36:3), PC (36:2), PC (36:1), PC (34:1e), PC (34:1), PC (32:1), PC (18:0/18:1), and PC (38:5), and the LPCs included LPC (18:1), LPC (18:0) and LPC (16:0). Conclusion Our findings indicate the presence of a comprehensive lipid profile in the vascular tissue of atherosclerotic mice, particularly involving PC and LPC, which exhibited significantly increased levels in AS. Supplementary Information The online version contains supplementary material available at 10.1186/s12986-022-00723-y.

show that lipid disorder is a principal determinant of the initiation and progression of AS [3]. Endogenous lipids are the major constituents of cell membranes, which are extensively involved in the structural compartmentalization of signal transduction. Moreover, these lipids act as potential pathophysiological mediators and participate in regulating several intracellular functions, such as proliferation, apoptosis, oxidative stress and inflammation [6]. Atherosclerosis is closely linked to lipid overload, including increased plasma levels of triglycerides (TG), low-density lipoprotein (LDL), intermediate-density lipoprotein (IDL) and VLDL [7,8]. For example, LDL, a particle that carries cholesterol through the blood, is not only the initiator but also an accelerator of AS [9,10]. Lipoprotein(a), a low-density lipoprotein (LDL) cholesterol-like particle that binds to apolipoprotein (A), has a potential causal relationship with atherosclerotic cardiovascular disease [11].
Lipidomics is a powerful analytical tool for analyzing the lipid composition in biological samples under physiological and pathological conditions to elucidate disease pathogenesis, identify biomarkers or study the effect and mechanism of drugs [12,13]. Liquid chromatography coupled to MS (LC-MS) is undoubtedly considered the current gold standard for the quantification of lipids [14]. Lipidomics has been widely used in screening and mechanistic research on lipid biomarkers in atherosclerotic diseases in recent years. For example, the plasma lipid levels in subjects with TG/HDL-C levels ≥ 3.287 were found to be significantly different from those in subjects with TG/HDL-C levels < 1.  (38:1), to the contrary, all TG species and phosphatidylserine (PS) (36:1) were increased [15]. In patients with large artery atherosclerotic cerebrovascular disease, 276 sphingolipids were detected in the plasma via lipidomic analysis, including SM (n = 162), glycosphingolipids (n = 72), Cer (n = 39), and ceramide phosphates (n = 3) [16]. Moreover, Wang et al. [17] identified nine types of 732 lipid components in the plasma of ApoE −/− mice, including 11 lysophosphatidylethanolamine (LPE), 12 SM, 15 LPC, 27 cholesteryl esters (CE), 35 ceramides (Cer), 38 diacylglycerol (DAG), 60 phosphatidylethanolamine (PE), 64 PC, and 470 triglyceride (TAG). Therefore, further comprehensive exploration of the lipid characteristics related to atherosclerosis, including in serum and vascular tissue, is conducive to enabling an in-depth understanding of the pathogenesis of atherosclerosis and is also expected to provide new insights for the diagnosis, prevention and cure of atherosclerosis.
In summary, omics approaches are used to identify lipid profiles in plasma, serum, and plaques to establish a disease prediction model, elucidate disease pathogenesis, improve diagnostic efficiency and provide drug targets. To date, due to the limited sample size, although Jihan Talib et al. have developed a new Multi-ABLE method that can be used to assess small tissue samples (< 5 mg) and have performed aortic lipidomics in ApoE −/− mice [18], there has been no literature describing the use of classic LC-MS techniques to assess the profile of vascular lipids in ApoE −/− mice. Therefore, we used classic LC-MS/MS lipidomics technology to provide a basis for elucidating the lipid profile of atherosclerotic thoracic aortas in ApoE −/− mice.

Animal model establishment and specimen acquisition
This study was approved by the Chengdu Women's and Children's Central Hospital Ethics Committee. ApoE −/− mice (4 weeks old, male) were purchased from Jiangsu Jicui Yaokang Biotechnology Co., Ltd. Animals were kept in a specific pathogen-free (SPF) environment. The feeding conditions included a light-dark cycle for 12 h, temperature 22-25 °C, and a relative humidity of 60%. The mice were divided into two groups: (1) mice were fed a normal diet, and (2) mice were fed a high-fat diet (Yangzhou Promoter Biotechnology Co., Ltd., D12108C). After feeding for 8 weeks, the mice were fasted overnight and anesthetized with sodium pentobarbital solution (1%, 10 mL/kg) by intraperitoneal injection. The plasma was separated from the blood, which was collected from the abdominal aorta, after standing for 2 h and centrifugation at 8000 rpm for 8 min. The plasma was transferred to an EP tube and preserved at − 80 °C refrigerator before being used for analysis. The thoracic aorta was peeled off and preserved at − 80 °C refrigerator.

The measurement of serum TC, TG and LDL levels
The cryopreserved plasma (100 μL) obtained following fasting was thawed and then centrifuged. The biochemical markers associated with lipid metabolism, including the levels of LDL, TC, and TG, were detected using a commercial kit according to the manufacturer's instructions (Changchun Huili Biotech Co., Ltd.).

Histopathologic analysis
The heart, including the aortic root, was fixed in paraformaldehyde. Then, the tissues were embedded in OCT compound and sliced into sections (7 μm) [19]. Then, HE and Oil Red O staining were performed to observe the plaque area and lipid deposition, respectively. Furthermore, Masson's trichrome staining was performed to observe the collagen area.
All images were photographed using a panoramic MIDI scanning microscope (3DHISTECH, Hungary) and then quantified using Image-Pro Plus 6.0 software. The plaque area was measured as the total area of plaque. Lipid deposition is presented as the percentage of positive cells in total plaque. The collagen area was measured as the collagen-positive area.

Immunostaining
As previously described, the expression of CD68 in the aortic root was assessed by immunostaining [20]. Briefly, the frozen sections were permeabilized with Triton X-100 (0.2%, 15 min) at room temperature. After incubation with rabbit anti-CD68 (1:100, Servicebio) at 4 °C overnight, the sections were washed, a secondary antibody was added, and the samples were incubated. Finally, DAPI was added to stain the nuclei. Images were acquired using a panoramic MIDI scanning microscope (3DHISTECH, Hungary) and then quantified using Image-Pro Plus 6.0 software.

Lipid extraction
Due to the limitation of sample weight, the thoracic aorta vessels of every two mice were combined in the analysis. After quickly freezing with liquid nitrogen, the samples were placed in and stored at − 80 °C refrigerator until analysis. Lipids were extracted as follows: (1) 20 μL Lyso PC-17:0 (0.1 mg/mL, dissolved in methanol) was added to the samples as internal standard to reflect the repeatability of the extraction. And then mixed the samples with methanol: water (1/1, vol/vol, 300 μL). Subsequently, the samples were subjected to grinding with two small steel balls (60 Hz, 2 min) in grinder (Tissuelyser-48, Shanghai) and then added 300 μL chloroform for ultrasonication for 10 min, and stored at − 20 °C for 20 min. After centrifugation for 10 min, 200 μL of the chloroform layer was transferred into a new centrifuge tube. (2) The residual samples were added to chloroform:methanol (2/1, vol/vol) and then vortexed for 30 s. The samples were ultrasonicated for 10 min, placed at − 20 °C for 20 min, and centrifuged for 10 min, and then taken 150 μL of the chloroform layer into the centrifuge tube. (3) The two chloroform layers were combined and then dried under a nitrogen stream. The samples were re-dissolved in 400 μL isopropanol:methanol (1/1, vol/vol), vortexed for 30 s and sonicated for 3 min. Before detection, the solution was filtered through a 0.22 μm organic phase pinhole filter. All sonication was performed in an ice-water bath, and centrifugation was performed at 13,000 rpm at 4 °C. QC samples were prepared by mixing aliquots of all samples, and then were injected at regular intervals (every 6 samples) into the samples throughout the analysis procedure to evaluate the stabilityof the equipment. All chemicals and solvents were analytical or HPLC grade.

Lipidomic profiling using LC-MS
Lipid profiling was performed using an UPLC system (Waters, Japan) coupled to a Q-Exactive plus mass spectrometer equipped with a heated electrospray ionization (ESI) source (Thermo Fisher Scientific, Waltham, MA, USA). In detail, the samples were separated using an ACQUITY UPLC BEH C8 column (100 mm × 2.1 mm, 1.7 um, Waters) and detected using ESI positive and ESI negative ion modes. During the analysis, the samples were kept at 4 °C. The gradient elution system consisted of two mobile phases, (A) acetonitrile: water (6:4, v:v, containing 10 mmol ammonium formate, 0.1% formic acid) and (B) isopropanol: acetonitrile (9:1, v:v, containing 10 mmol ammonium formate, 0.1% formic acid). The flow rate was 0.26 mL/min, and the column temperature was 55 °C. The injection volumes were 2 μL and 4 μL in positive and negative modes, respectively. The separation was achieved using the following gradient: 0 min, 32% B; 1.

IPA analysis
Ingenuity pathway analysis (IPA) is a widely used bioinformatics software for predicting the correlations among components, targets, pathways, and diseases [21,22]. The HMDB number of differential lipids was downloaded from the HMDB database (https:// hmdb. ca/) [23]. In canonical pathway analysis and analyses of disease and function, we took − log (P value) > 2 as the threshold and then defined Z score > 2 and Z score < − 2 as the threshold of significant activation and inhibition, respectively. The consistency scores were calculated by assessing regulatory effects and molecular networks, and the higher the consistency score was, the more accurate the regulatory effect analysis results. In addition, we used a P value of overlap < 0.05 as the inclusion criteria for upstream regulators.

Statistical analysis
The original data obtained from Q Exactive plus (LC-MS/MS) analysis were processed by Lipid Search software. The molecular structure of lipids and the additive mode of its positive and negative ions were confirmed according to the parent ions and multistage mass spectrometry data of each individual sample. To sort the original data matrix, results were aligned according to a certain retention time range. The peak signals in each sample were normalized, by converting its signal intensity into the relative intensity in the spectrum, and multiplying it by 100,000. Lipids which the relative standard deviation (RSD) of QC greater than 30% were discarded. These highly variable ions should be rejected as unacceptable for the analysis. Following the extraction of data, peaks with missing values (ion intensity = 0) in more than 50% of groups were removed and replaced the zero value with half of the minimum value. Positive and negative ion data were combined into a data matrix. Principal component analysis (PCA) was carried out to monitor the overall sample distribution and the stability of the whole analysis process. Orthogonal partial least squaresdiscriminant analysis (OPLS-DA) and partial least squares-discriminant analysis (PLS-DA) were used to authenticate the results regarding the differential lipids. Variable importance of projection (VIP) value was used to assess the variable importance. In addition, the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway database was used to analyze metabolic pathway enrichment. The statistical graphs were generated by Graph-Pad Prism 8 software, and a two-tailed Student's T test was used for statistical analysis. All data are presented as the means ± standard errors of the means, SEM). Differential lipids were selected with VIP values > 1.0 and p values < 0.05.

Effects of a high-fat diet on the lipid profile in ApoE −/− mice
We used male ApoE −/− mice fed a high-fat diet to establish an animal model of atherosclerosis and then collected blood vessel tissue to extract lipids for mass spectrometry analysis. The study design is shown in Fig. 1A. In brief, ApoE −/− mice were fed a high-fat diet (HFD) for 8 weeks to establish an atherosclerosis model, while ApoE −/− mice fed a normal diet (ND) were used as a control. The lipid species in thoracic aorta were assessed via HPLC-MS/MS. The body weight was not significantly different between the ND group and the HFD group (Fig. 1B). The plasma levels of TC, LDL-C, and TG were detected to investigate the effects of HFD on total lipids. The levels of TC and LDL-C in the HFD group were significantly higher than those in the ND group, while no significant differences in TG levels were observed; Fig. 1C-E illustrates the details.

Assessment of atherosclerotic plaque formation in ApoE −/− mice
To verify that the model of atherosclerotic mice was successful, HE and Oil Red O staining were performed in frozen sections of the aortic root of ApoE −/− mice. The results of HE staining showed that 8 weeks of HFD treatment significantly increased plaque formation ( Fig. 2A, B). Oil Red O staining showed that lipid deposition was also significantly increased in HFD-treated ApoE −/− mice (Fig. 2C, D). Masson's trichrome staining showed that the collagen content was significantly increased in HFD-treated ApoE −/− mice (Fig. 2E, F). In addition, the immunofluorescence staining of CD68 expression revealed that free monocyte/macrophage levels in the aortic root were increased (Fig. 2G, H). The above results revealed that ApoE −/− mice treated with HFD did indeed undergo the pathological changes associated with atherosclerosis, which suggested the success of animal model establishment.

UHPLC-Q Exactive MS-based lipidomics analysis
UHPLC-Q Exactive Plus-MS was used to lipidomic analysis of the thoracic aorta samples. The molecular structure of lipids and the additive mode of the positive and negative ions were identified according to the parent ions and multistage mass spectrometry data obtained from each individual sample. The base peak chromatograms (BPCs) of positive ions and negative ions in the ND group and HFD group are shown in Additional file 1: Fig. S1, respectively.
To evaluate the stability of the equipment, the lipid intensity distribution of the QC samples is shown in Fig. 3A, where the Y coordinate is the log10 value of the mass spectrum intensity. PCA was used first to monitor the overall distribution of the samples and the stability of the entire analysis process, as shown in (Additional file 2: Fig. S2). The distance between the two coordinate points on the score chart was far on the score map, indicating that there was a significant difference between the two samples. PLS-DA and OPLS-DA were used to identify differential lipids between groups (Fig. 3B, Additional file 2: Fig. S2). A random ranking method, the response ranking test, was used to evaluate the accuracy of the OPLS model. The value (green) of all Q2 on the left was lower than the initial value on the right, which shows that the model was effective (Fig. 3C). An S-plot showed that all the points were distributed in the first and third quadrants, and the closer the lipid molecules were to the lower left or upper right of the plot, the more significant the differences were (Fig. 3D).

Analysis of lipidomic profiling and screening of differential lipids
Based on the PLS-DA model, a total of 131 differential lipids were identified according to the conditions of VIP > 1 and P < 0.05. A total of 57 lipids exhibited increased levels, and 74 exhibited decreased levels in the HFD group (Fig. 4A). The details of the differential lipids are shown in Table 1 Table 2. We performed Z score analysis on the top 20 significantly different lipids sorted by VIP and transformed the data of different magnitudes into a unified measure of Z score for comparison (Fig. 4D). To study the degree of correlation . E Measurement of TG content in the plasma of mice (n = 10 per group). All data are presented as the means ± SEMs. ***P < 0.001 versus the ND group. ns, not statistically significant. LC/MS, liquid chromatography-mass spectrometry; TC, total cholesterol; LDL, low-density lipoprotein; TG, triglyceride; ND, normal diet; HFD, high-fat diet between lipid levels, we performed correlation analysis using the top 20 significantly different lipids sorted by VIP and used the Pearson correlation coefficient to measure the degree of linear correlation between lipid levels. A positive correlation is indicated in red, while a negative correlation is indicated in blue. The larger the dot is, the larger the correlation coefficient between the two variables (Fig. 4E). In addition, we selected the relationship pair with p value < 0.05 and correlation > 0.95 to draw a correlation network diagram (Fig. 4F).

Metabolic pathway enrichment analysis
KEGG pathway enrichment analysis (http:// www. kegg. jp/ or http:// www. genome. jp/ kegg/) was used to predict metabolic pathways based on the lipids with increased and decreased levels. As shown in Fig. 5A, the lipids with increased levels were mainly associated with glycerophospholipid metabolism, choline metabolism, retrograde endocannabinoid signaling, autophagy-other, systemic lupus erythematosus, autophagy-animal and cholesterol metabolism. Interestingly, the lipids with decreased levels were mainly associated with fat digestion and absorption, regulation of lipolysis in adipocytes, retrograde endocannabinoid signaling, insulin resistance, . All data are presented as the means ± SEMs. ***P < 0.001 versus the ND group. HE, hematoxylin-eosin; ND, normal diet; HFD, high-fat diet thermogenesis, glycerophospholipid metabolism and Th1 and Th2 cells differentiation (Fig. 5B). These results indicate that lipid metabolism imbalance, autophagy and inflammatory responses may be involved in the mechanism underlying atherosclerosis. The above results showed that glycerophospholipid metabolism was significantly enriched in both the activated and inhibited KEGG pathways, which attracted our attention. Further analysis of individual glycerophospholipid pathways revealed that some lipids, including phosphatidyl-Lserine, phosphatidyl-ethanolamine, phosphatidylcholine (Lecithin) and 1-acyl-sn-glycero-3-phosphocholine, exhibited increased levels, which may play a critical function in the occurrence or progression of AS (Fig. 5C). Moreover, IPA network pathway analysis showed that differentially expressed lipids were closely related to NF-κB signaling, PI3K/AKT signaling and JAK/STAT signaling (Additional file 3: Fig. S3).

Lipid composition analysis
The  Fig. S4). Interestingly, the levels of all lipids in the PC and LPC classes were significantly increased in the HDF group, which may play a nonnegligible effect in the onset and process of atherosclerosis.
As reported, PCs [24,25] and LPCs [26] are closely related to atherosclerosis. To explore the role of PC and LPC in atherosclerosis, we further analyzed the expression levels of 10 PCs and 3 LPCs, which exhibited significantly increased levels in the HFD group.

Discussion
Dyslipidemia is a major risk factor for atherosclerosis and atherosclerosis-related cardiovascular disease [27]. A consensus has been reached on the conclusion that elevated plasma LDL-cholesterol levels promote AS. LDL lipoproteins penetrate damaged endothelial cells, accumulate in the intima after oxidation, and subsequently recruit monocytes to endothelial cells, causing monocytes to differentiate into macrophages and form foam cells, initiating and accelerating the process of atherosclerosis [9]. Unlike cholesterol, triglycerides are also associated with subclinical atherosclerosis and vascular inflammation [28]. In this study, the lipid profile of the blood vessels in AS mice was investigated through lipidomic technology, and 131 differential lipids were identified in the thoracic aorta of ND-and HFD-treated Recently, omics approaches have been used to improve the understanding of the pathogenesis of AS. For example, Wang et al. [17] used ultraperformance liquid chromatography coupled with quadrupole time-of-flight    Recent studies have shown that glycerol phospholipid metabolism plays a critical role in the pathogenesis of AS, and disordered glycerol phospholipid metabolism directly affects the process of AS [29,30]. In our study, the KEGG enrichment analysis of the lipids that were identified to exhibit increased and decreased levels revealed enrichment in glycerol phospholipid metabolism, indicating that glycerol phospholipid metabolism may be an initiating factor; these results are conducive to further elucidating the pathogenesis of AS. As reported, the major classes of glycosphingolipids, comprising phosphatidylcholine (PC), PE, and PS, are closely related to atherosclerosis. In all nucleated mammalian cells, PC and PE are the most abundant phospholipids in cell membranes [31,32]. PC is mainly synthesized by the CDP-choline pathway [33]. In an additional pathway, under the condition of S-adenosylmethionine (AdoMet) as a methyl donor, PE can synthesize PC through three sequential methylation reactions [34]. Impaired phosphatidylcholine biosynthesis has been reported to reduce atherosclerosis in ApoE −/− mice [24]. Exteriorized PS and PE residues on the cell membrane in apoptotic cells trigger rapid phagocytosis and contribute to plaque vulnerability [35][36][37]. Among patients with CAD, SMs are directly related to serum levels of total cholesterol and lipoproteins, which suggests that sphingolipids play a critical role in modulating lipid serum levels. The results above highlight the importance of establishing new goals in the glycerol metabolic pathway for antiatherogenic therapies [38].
In addition, we found that the levels of 10 PCs from HFD-treated ApoE −/− mice were significantly higher than those in the ND group, including PC ( (38:5). Phosphatidylcholine (PC) is a subclass of glycerophospholipids with choline as the head group. The physiological functions of PC include membrane integrity/stability, cell proliferation and survival, fuel and energy storage/ source, glycerophospholipid metabolism and cell signaling [31,39]. It has been reported that the levels of PCs (16:2/22:6, 16:2/20:5, 15:0/18:2) were decreased in stroke patients [40]. In contrast, PC (16:0/20:4) levels were significantly higher in the serum of patients with severe calcific coronary artery disease than in patients with noncalcific coronary artery disease, while PC(18:2/18:2) and PC(36:3) levels had the opposite trend [41].
Lysophosphatidylcholine (LPC) is the main component of oxidized low-density lipoprotein (oxLDL). In cells, saturated LPC (16:0) and PC can be interconverted. Specifically, PC can be cleaved by phospholipase A2 (PLA2) to form LPC [42,43], and LPC can be converted to PC by lysophosphatidylcholine acyltransferase (LPCAT) [44]. In addition, lecithin-cholesterol acyltransferase (LCAT) can transfer fatty acids to free cholesterol to form LPC [44]. LPC damages endothelial cells, induces smooth muscle cell proliferation and lymphocyte and macrophage migration, increases inflammatory mediator production and oxidative stress, and induces cell apoptosis and necrosis, which accelerate the progression of atherosclerosis and atherosclerosis-related cardiovascular diseases [26,45,46]. In patients with hypertension, LPCs can be considered potential predictors for further evaluation and validation of early vascular aging (EVA) [47]. As reported, the levels of lysoPC 16:0, 18:0, and 18:1 in human plaques are markedly related to inflammatory mediators, such as monocyte chemoattractant protein-1 (MCP1), interleukin-1β (IL-1β), interleukin-6 (IL-6), and macrophage inflammatory protein-1β (MIP-1β) [48], which suggests that lysoPCs play a key role in plaque inflammation and stability. In our study, we also found that the levels of these three lipids were elevated in the vascular tissue of mice with atherosclerosis, which was consistent with the above research report. This finding indicates that LPC (16:0), LPC (18:0) and LPC (18:1) may participate in the process of atherosclerosis and be closely related to vascular injury.
Although we revealed the lipidomic profiling of vascular tissue of atherosclerotic mice by LC-MS/MS technology and identified some differentially expressed lipid molecules, there are still some limitations. First, due to the continuous development of LC-MS/MS technology, lipidomics methods are widely used to comprehensively analyze lipids and their interactions with other molecules. However, it is not clear whether all lipids in some biological samples can be comprehensively detected and accurately identified due to the complex composition of lipids. It is also difficult to conduct accurate localization analysis on the detected lipids. Therefore, the development of new instruments and new bioinformatics schemes, such as ion mobility mass spectrometry (IM-MS) [49], NanoLC-MS [50], AFADESI-MSI [51]and spatial metabolomics analysis [52], may bring new insights into the application of metabolomics and lipidomics in the future. Second, our study suggested that PCs and LPCs may have a strong association with atherosclerosis, but the causal relationship between them needs to be further clarified. In addition, we observed an interesting phenomenon. While confirming the successful establishment of the atherosclerosis model, there was no significant difference in the TG levels in the serum between mice in the HFD group and those in the ND group. More interestingly, our lipidomics spectrum analysis also revealed that the levels of TG lipid molecules were not increased in the HFD group, and the levels were in fact lower. It is possible that the specificity of TG as a marker of AS needs further reference.

Conclusion
Our findings indicate that in the comprehensive lipid profile in the vascular tissue of atherosclerotic mice, a total of 131 differential lipids were identified, including 57 lipids with levels that were increased in the HFD group and 74 with levels that were decreased. Further analysis revealed that the identified differentially expressed PCs and LPCs exhibited significantly increased levels. We particularly analyzed the details of PCs and LPCs, which exhibited significantly increased levels in AS. Therefore, targeting PCs, LPCs and lipid metabolism might be a novel and effective therapeutic strategy for atherosclerosis.