Proteomic analysis of postprandial high-density lipoproteins in healthy subjects

Depósito de investigación de la Universidad de Sevilla https://idus.us.es/ “This is an Accepted Manuscript of an article published by Elsevier in: INTERNATIONAL JOURNAL OF BIOLOGICAL MACROMOLECULES on 2023, available at: https://doi.org/10.1016/j.ijbiomac.2022.11.187” Proteomic analysis of postprandial high-density lipoproteins in healthy subjects Elena Grao-Crucesa,†, Alejandro Santos-Mejiasa,†, Ignacio Orteab, Elvira Marquez-Paradasa, Maria E. Martinc, Sergio Barrientosd, Beatriz Bermudezc, Sergio Montserrat-de la Paza,* a Department of Medical Biochemistry, Molecular Biology, and Immunology, School of Medicine, University of Seville. Av. Sanchez Pizjuan s/n, 41009 Seville, Spain b Proteomics Unit (CINN-CSIC), Health Research Institute of Asturias (ISPA), Av. Hospital Universitario s/n, 33011 Oviedo, Spain c Department of Cell Biology, Faculty of Biology, University of Seville. Av. Reina Mercedes s/n, 41012 Seville, Spain d Department of Nursing, Faculty of Nursing, Physiotherapy and Podiatry, University of Seville. Avenzoar st., 41009 Seville, Spain † Elena Grao-Cruces and Alejandro Santos-Mejias should be considered joint first author *Corresponding Author: Prof. Sergio Montserrat-de la Paz Department of Medical Biochemistry, Molecular Biology, and Immunology School of Medicine, Universidad de Sevilla, Av. Sanchez Pizjuan s/n, 41009 Seville (Spain) E-mail: [email protected] Manuscript File (Word Document ONLY) Click here to view linked References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Running title: Proteome of postprandial HDL Conflicts of interest: The authors state no conflict of interest. Funding: This study was supported by the research Grant US-1263458 (Andalusian Ministry of Economy, Knowledge, Business, and University, Government of Andalusia, Spain) into the European Regional Development Fund Operational Programme 2014 to 2020. Acknowledgements: Elena Grao-Cruces and Elvira Marquez-Paradas have the benefit of a doctoral fellowship (BES-2012-056104) from the Andalusian Government and a departmental collaboration grant awarded by the Ministry of Education and Vocational Training from the Spanish Government, respectively. The authors thank Carlos Fuentes from the Proteomics Unit of the Central Research Support Service of Cordoba University (UCO) for its technical assistance during the fulfilment of this work. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 ABSTRACT The relationship between the functionality and composition of high-density lipoproteins (HDL) is yet not fully studied, and little is known about the influence of the diet in HDL proteome. Therefore, the aim of this research was to elucidate the HDL proteome associated to postprandial hyperlipidemia. Male volunteers were recruited for an interventional study with high fatty acid-based meals. Blood samples were collected before the intake (baseline), and 2-3 (postprandial peak) and 5-6 (postprandial post peak) hours later. HDL were purified and the protein composition was quantified by LC-MS/MS. Statistical analysis was performed by lineal models (amica) and by ANOVA and multi-t test of the different conditions (MetaboAnalyst). Additionally, a clustering of the expression profiles of each protein was done with coseq R package (RStudio). Initially, 320 proteins were identified but only 119 remained after the filtering. APOM, APOE, APOB, and APOA2, proteins previously identified in the HDL proteome, were the only proteins with a statistically significant altered expression in postprandial hyperlipidemia when compared to baseline (p values < 0.05 and logFC > 1). In conclusion, we have been able to describe several behaviors of the whole HDL proteome during the postprandial hyperlipidemic metabolism. Keywords: High-density lipoprotein; proteome; hyperlipidemia; postprandial metabolism; apolipoprotein. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 1. Introduction High-density lipoproteins (HDL) are a complex and diverse group of blood circulating biomolecules, their initial structure is configurated in a discoidal shape, with a low proportion of lipids, that maturates to a spherical one due to the accumulation of cholesteryl ester (CE) inside it, which is the main way cholesterol is associated with HDL [1]. The general structure of HDL particles consists of a lipid core, mostly CE and triacylglycerides (TAG); and an outer layer made of phospholipids and proteins, primarily APOA1 and APOA2 [2]. However, HDL particles are an incredibly heterogenous group of protein-lipid complexes that vary in size, shape, density, and composition, they can carry more than hundreds of different proteins and lipids, differing highly in their composition [3-5]. Originally, HDL were defined in 5 subclasses by ultracentrifugation density [6]. Nevertheless, HDL definition is slowly turning over the year to a protein-based standard, initially separated in 3 classes depending on APOA1 and APOA2 presence [7], but currently 16 unique HDL particle types has been proposed based on their protein composition [8]. The way of defining HDL influences the method for isolation, which is one of the key factors in the detection of HDL- associated proteins. The most used isolation method is ultracentrifugation, but immunoaffinity gels are also an option [5]. Additionally, considering that the other key factor in the protein detection is the mass spectrometry technology used [9], it is crystal-clear that detection capability of HDL proteins varies immensely depending on the experimental design. Nonetheless, not only experimental design has to be taken into account, environmental factors, such 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 as metabolic alterations, pathogeny, and diet of the subjects; have shown to influenced HDL composition [10-12]. This heterogeneity within HDL composition affects their functionally [4]. It has been shown by several research groups that the functional state of HDL particles is highly driven by their protein composition [4,11,13,14]. Thus, a better understanding of this relationship is a matter of relevancy for their therapeutic implications. For instance, the inverse relation between HDL-cholesterol and cardiovascular diseases is well-accepted in the scientific community, but strategies targeting to raise HDL-cholesterol has been unsuccessful to lower cardiovascular diseases risk [3,4,11,15-18]. Furthermore, alternative strategies targeting HDL composition has revealed themselves to be extremely promising for a better diagnosis and treatment of cardiovascular diseases [11,12,18]. In fact, due to HDL nature as circulating particle transporter, they are implicated in many biological processes, namely, activation of the complement pathways, modulation of inflammation process, control of the hemostasis, and some cell binding activities [19]. But essentially, they are key for the transportation of cholesterol from peripherical areas to the liver to be catabolized, and HDL particles undergo a continuous remodeling, due to the interaction with other circulating particle through this process [20]. As previously stated, it is not fully understood the relation between protein composition and functionality, and little is known about the influence of the diet in HDL proteome. Actually, dietary intervention studies aimed to analyze HDL proteome are scattered and mostly focused on the characterization of HDL functionality, not in biomarker identification. In addition, there is a lacking in solid experimental conditions due to the difficulties of this kind of interventions. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Also, there is just a few research following an omics strategy, with most of the studies focusing on several known HDL-associated proteins to measure HDL composition changes rather than identification of new proteins. Nonetheless, they have shown that HDL proteome is modified by the type of diet [12, 21-26]. Besides the limitations, it seems that there are certain changes in particular HDL proteins, but the conclusions are problematic due to the lack of homogenous conditions. While Morgantini et al. [10] found an increase in APOB and APOA1 in young volunteers that underwent a high-fat diet for 14 days, Padró et al. [22] did not appreciate this change in APOA1, however they found a significant change in APOM, RBP4, and CRABP-I. Finally, the only research focused on hyperlipidemia during the postprandial state, which is the physiological risen of plasma lipids after a meal, did not find any change in APOA1 and APOB after the ingestion of high saturated fat meals, rather it depicted a change in total protein and lipid composition in healthy adult subjects [25]. Therefore, the aim of this study is to elucidate HDL proteome associated to postprandial hyperlipidemia, which is induced by an intake of high-fatty acid emulsions in healthy adult male volunteers. So far in this research, we have been able to identify 277 proteins, with 119 remaining after filtering out irrelevant ones. Four proteins showed a statistically significant altered expression: APOB expression seems to be altered during the postprandial peak and APOA2, APOM, and APOE during the postprandial post peak. Finally, we have been able to describe several behaviors of the whole HDL proteome during the postprandial hyperlipidemic state, finding relevant ones being related to immune system modulation and transport of biomolecules. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 2. Materials and Methods 2.1. Patient cohort and study design This study was conducted according to Good Clinical Practice Guidelines and in line with the principles outlined in the Helsinki Declaration of the World Medical Association. Ethics approval was obtained from the Human Clinical Research and Ethics Committee of the University Hospital Virgen Macarena (PI00082017) and all subjects gave written informed consent. Adult male volunteers aged 25 to 35 years, non-smokers, with no medical history of disease known, abnormality of hematological or biochemical parameters were recruited in the Laboratory of ImmunoNutrition and ImmunoMetabolism at the School of Medicine (University of Seville). After an overnight fasting period of 12 h, all of them were given an oral fat emulsion prepared according to the method described by our Patent WO/2014/191597. They consisted of water, sucrose, fat (50 g/m2 body surface area), emulsifier, and flavoring. At fasting (baseline) and after the ingestion of the meals within 10 min, blood samples were collected 2-3 (postprandial peak) and 5-6 (postprandial post peak) hours later. Blood was collected into K2EDTA (Becton-Dickinson, San Jose, USA), the samples were centrifuged (140 g, 10 min at 4°C), and plasma was frozen at -20 ºC until HDL purification analysis. HDL were purified from plasma using a commercial kit (STA-608, CellBiolabs INC, San Diego, USA) according to the manufacturer's instructions. 2.2. Label-free proteomic quantification 2.2.1. Sample preparation for mass spectrometry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 HDL samples were processed and analyzed by the Proteomics Unit of the Central Research Support Service of Cordoba University. Samples were prepared for mass spectrometry (MS). Firstly, quantification of the total protein amount by the Bradford method was performed. Samples underwent through an SDS-PAGE electrophoresis gel to equalize protein quantity for digestion. Only 30 µg of proteins were digested with trypsin. Digestion was stopped using 0.5% Trifluoroacetic acid (TFA). Peptides were desalted using a Pierce C18 Spin Columns (Thermo Scientific) according to manufacturer’s instructions. Finally, peptides were concentrated using a SpeedVac and resuspended in 30 µL of mobile phase (2% Acetonitrile, 0.05% TFA) before being injected into liquid chromatography-mass spectrometry (LC-MS)/MS. 2.2.2. High-performance Liquid Chromatography-Mass Spectrometry/tandem Mass Spectrometry (HPLC-MS/MS) Resuspended peptides were separated in an Ultimate 3000 Dionex UHPLC (Thermo Scientific). Samples were loaded into a C18 pre-column (PepMap100, 5 µm, 300 µm x 5mm, Thermo Scientific), and then a 5 µL acetonitrile gradient from 4 to 40% was applied at a flow rate of 300 nL/min in a C18 nano-column C18 (Acclaim PepMap RSLC 75 µm x 50 cm, Thermo Scientific) for 60 min. The total chromatography time was 85 min. An Orbitrap Fusion (Thermo Scientific) in Data Dependent Acquisition double mode, equipped with a nano electrospray ionization ion source was used for MS. The Orbitrap Fusion worked in a positive voltage of 2 kV. Full scan mass range was 400-1500 m/z with a resolution of 120K, AGC values of 4 x 105 and a maximum injection time of 50 ms. Peptide analysis was performed in data dependent acquisition mode operating in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 surfaces. Furthermore, it is the main ligand for LDL receptor [50]. APOE decreasing during the postprandial post peak is well-known phenomenon [51]. As a matter of fact, its capacity to modulate APOB-rich lipoprotein clearance is partly what induces its content reduction in the postprandial state [52-55]. Additionally, APOE influences the size capability of HDL, allowing those particles to increase in size and storing more TAG/cholesterol [1]. Taken together, it looks like APOE takes a fundamental role in lipid metabolism, strengthening their relevancy during hyperlipidemia state. Interestingly, APOE has also being relevant in the neurobiology field due to their capability of regulating lipid metabolism and inflammation in the brain, and Alzheimer studies have frequently focused on APOE [50,56-58]. In brief, it is coherent, and consistent with their known functionality, that all the mentioned proteins have an altered expression during the hyperlipidemia state, as all of them participate in one way or another in lipid metabolism, and in the case of APOB is a consequence of the regulation of this biological process. 3.3. Functional enrichment analysis supports the role of APOB, APOM, APOA2, and APOE in lipid metabolism Functional enrichment analysis was performed to check whether APOB, APOM, APOA2, and APOE were, firstly, HDL-related proteins and, finally, their relevancy in the lipid metabolism. The enriched terms in the Gene ontology (GO) cellular component database (Figure 4A) support that our significant proteins belong to lipoprotein particles. Furthermore, all of them are recorded to be identified in HDL related research several times in the “HDL Proteome Watch” database [5]. Thus, it can be assumed with high level of certainty that 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 our significant proteins are HDL-associated proteins. Regarding the GO molecular function database (Figure 4B), it can be concluded they are implicated mostly in lipoprotein remodeling and modulation of their metabolism, enriched functional terms such as lipid transport activity, lipoprotein receptor binding, and lipid transfer activity. This suggests their role in the postprandial response is focused, as mentioned in the previous section, essentially in allowing the lipid transport and increasing the stability of HDL particles during the frequent interaction with APOB-rich lipoprotein, with the aim to work efficiently in the clearance of TAG in circulation. Additionally, most of the biological processes in which they are involved are related to the modulation of lipid homeostasis, HDL structure, and their metabolism (Figure 4C). Accordingly with GO databases findings, most of the enriched pathways are related to lipoprotein remodeling and assembly (Figure 4D). Once more supporting their fundamental role in stabilizing HDL particles during the postprandial process. Surprisingly, all of them are also implicated in non-related lipid metabolism pathways, namely, the sensory perception and visual phototransduction, probably due to their pleiotropic capabilities. In conclusion, the enrichment functional analysis suggests the role of APOB, APOM, APOE, and APOA2 in the postprandial response is due to their capability of modulating HDL remodeling and enhancing lipid transportation. This may be facilitated by providing greater stability to HDL particle in a more lipid content context (APOA2) or by helping in the increase interaction between HDL to APOB-rich lipoproteins (APOM, APOB, and APOE). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 3.4. Co-expression clustering analysis shows relevant expression patterns of proteins that modulate non-lipid metabolism functions In addition to the differential expression analysis, we performed a co-expression pattern clustering with the preprocessed dataset. With this kind of analysis, we aimed to identify similar behavior between proteins, which, although not differentially expressed; were present during the postprandial process. We achieved to classify all the proteins in 8 clusters (Figure 5). Nevertheless, cluster 4, 6, and 8 were totally irrelevant partly because of their size and chaotic behavior, cluster 4 consists just in 1 protein with an incomprehensible behavior, meanwhile cluster 6 and 8 have a subject-dependent profile, which proteins of those clusters only expressing in one of the volunteers nor in the others. Cluster size can be checked in Table 1. Focusing on the relevant clusters, those which bring some insight on the phenomenon, there were three patterns found (Figure 6): increasing, decreasing or no change in the expression through the whole timepoints. Clusters with a general increase of expression along their members are cluster 1 and 3, the latter in a lighter manner than the first one, cluster 1 contains APOB. Cluster 5 consists of proteins with almost no change along the whole process, so in this group we could find all the HDL proteins not affected by hyperlipidemia. Lastly, cluster 2 and 7 proteins decrease their expression along the time. From now on, we will refer to these two clusters as cluster D, this decision was based in two key points: firstly, consistency to keep all the significant proteins from the differential expression analysis in the same cluster (APOE and APOA2 belong to cluster 2, while APOM to cluster 7); and lastly, the enormous similarity in the behavior. The expression pattern of the significant differentially expressed proteins is consistent with the results of the 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 differential expression analysis. Regarding the possible functionality of each cluster, it seems cluster 1 only contains proteins related to activation cascade pathway and immune system, more than half of the proteins in this cluster participate in those pathways (Figure 7A). Similarly, cluster 3 have a mixed functionality, the pathways modulated by the highest proportion of proteins in this cluster are those involving hemostasis and platelet capability modulation, in a minor manner this cluster is also implicated in immune system and complement cascade pathway regulation. Interestingly cluster 3 shares functionality in a major manner with cluster D (2+7) and in a minor manner with cluster 1. Both cluster D and 3 are involved in hemostasis and platelet regulation (red boxes in Figure 7B), but they have an inverse expression pattern, suggesting along the process these two clusters might be acting coordinately to regulate hemostasis through platelet regulation. However, it seems that proteins in cluster D are also implicated in HDL remodeling. In fact, APOE, APOM, and APOA2 belong to this cluster, so proteins in this cluster are potentially in charge of the modulation in lipid metabolism itself. Functions shared between cluster 1 and 3 are those related to complement and coagulation cascade pathway modulations, they both have an increase of expression along the time pattern. Nevertheless, complement cascade modulation seems to be influenced by proteins in all the relevant clusters, suggesting this pathway is implicated in the postprandial response, inhibiting some of its roles by cluster D proteins, while activating some others by cluster 1 and 3 (Green box in Figure 7). The anti-inflammatory capacity of HDL is wellknown, this capacity in our case could be modulated probably through the complement cascade related proteins [18,59,60]. Finally, cluster 5, the cluster 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 with no change in expression, is involved in several pathways, mostly immune system modulation through the complement cascade and hemostasis through platelet regulation, pathways also modulated by the other clusters. The level of activation/deactivation of those two pathways is unclear and escape the aim of this research work, further efforts are needed to elucidate HDL proteome implications in immune system and hemostasis during postprandial hyperlipidemia. 3.5. Protein-protein interaction (PPI) network analysis Regarding the results obtained from the PPI network analysis, there are some remarks to highlight. On the one hand, from the differential expression analysis, APOB, APOM, APOE, and APOA2 are well-known interacting proteins. They create a non-random network with high confidence connections between all of them (>0.9) (Supplementary Figure 1). These findings support further what has already been said above, most likely these four proteins work coordinately to modulate postprandial hyperlipidemia response. On the other hand, PPI network analysis of the different co-expression clusters yield similar results, all of them being non-random networks with high interaction between their nodes. In cluster 1 network, there is known interactions between almost all proteins, and even though some of those relationships are weaker than others, there seem to be a solid central cluster (red shape in Supplementary Figure 2). In cluster 3 network, it looks like there are two groups of proteins linked to each other, the largest groups could be those proteins associated to hemostasis, as there are several proteins involved in this pathway included in it, such as SERPING1, SERPINF2 and SERPINC1 (red shape in Supplementary Figure 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 3). However, it is not clear how the smallest group, which consists mainly in cytoskeleton-related proteins, could participate in the functions associated to this cluster (increase signaling of Hemostasis/Immune system modulation along the postprandial hyperlipidemia) (green shape in Supplementary Figure 3). The network of cluster D is similar in shape to cluster 1 network, as there is only one large cluster of high interconnected proteins, which is made of apolipoproteins (red shape in Supplementary Figure 4). It is coherent that most of known relationships in this network are carried by lipoproteins, as they made up more of the half of the cluster and it is the only cluster directly focused on lipoprotein remodeling and lipid metabolism. Finally, regarding cluster 5 network, the confidence cutoff had to be increase to 0.7, meaning only high confidence interaction are shown, because of the high number of proteins in this cluster. This cluster does not change its expression level along the whole process. Here, there are four groups of proteins bound by linking proteins, such as APOA4, C8A and PLG. Most of its proteins are implicated in modulation of immune system through complement cascade pathway regulation, which is sound with the large number of complement cascade proteins in the network (blue shape in Figure 8); and regulation of hemostasis through platelet activity modulation, the rest of proteins might be involved in this issue. In fact, the group of proteins marked in red and orange are known to be involve in platelet activity, the former modulating different conditions of plasma; and the latter through cytoskeletal binding activity regulation (red and orange shape in Figure 8). Surprisingly, there are a few HDL-associated proteins that group together, it is not clear how they are connected to hemostasis group (green shape in Figure 8). What this cluster 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 and its network make clear is that even though hemostasis, immune system and lipoprotein remodeling pathways are affected by other clusters, they do it in a precise and partial way as most of the HDL proteins are connected to each other. In conclusion, the PPI network analysis show that almost all the proteins in our results firstly, in the differential expression analysis, and lastly, in the clustering analysis, are indeed interconnected to each other, adding reliability to our conclusions. 4. Conclusions Our findings point out that HDL proteome is influenced by postprandial hyperlipidemia, increasing the abundance in certain moments of the process of lipid metabolism and lipoprotein remodeling related proteins (APOB, APOM, APOE, and APOA2). In addition, expression patterns observed suggest that dietary fatty acids could also modulate other physiological functions, such as the immune system response, through HDL composition. Author Contributions Conceptualization, A.S.-M., I.O., and S.M.-d.l.P.; methodology, A.S.-M., E.G.- C., and I.O.; formal analysis, A.S.-M., and E.M.-P.; investigation, A.S.-M., E.G.- C., S.B., and M.E.M.; writing—original draft preparation, A.S.-M.; writing— review and editing, S.M.-d.l.P.; funding acquisition, B.B. and S.M.-d.l.P. All authors have read and agreed to the published version of the manuscript. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 References [1] Ong KL, Cochran BJ, Manandhar B, Thomas S, Rye KA. HDL maturation and remodelling. Biochim Biophys Acta Mol Cell Biol Lipids. 2022;1867:159119. [2] Ben-Aicha S, Badimon L, Vilahur G. Advances in HDL: Much More than Lipid Transporters. Int J Mol Sci. 2020;21:732. [3] Annema W, Von Eckardstein A. High-density lipoproteins - Multifunctional but vulnerable protections from atherosclerosis. Circ J. 2013;77:2432–448. [4] Cardner M, Yalcinkaya M, Goetze S, Luca E, Balaz M, Hunjadi M, et al. Structure-function relationships of HDL in diabetes and coronary heart disease. JCI Insight. 2020;5:e131491. [5] Davidson WS, Shah AS, Sexmith H, Gordon SM. The HDL Proteome Watch: Compilation of studies leads to new insights on HDL function. Biochim Biophys Acta Mol Cell Biol Lipids. 2022;1867:159072. [6] Havel RJ, Eder, HA, Bragdon, JH. The distribution and chemical composition of ultracentrifugally separated lipoproteins in human serum. J Clin Invest. 1955;34:1345–53. [7] Alaupovic P. The concept of apolipoprotein-defined lipoprotein families and its clinical significance. Curr Atheroscler Reports. 2003;5:459–67. [8] Furtado JD, Yamamoto R, Melchior JT, Andraski AB, Gamez-Guerrero M, Mulcahy P, et al. Distinct Proteomic Signatures in 16 HDL (High-Density Lipoprotein) Subspecies. Arterioscler Thromb Vasc Biol. 2018;38:2827–42. [9] Silva ARM, Toyoshima MTK, Passarelli M, Di Mascio P, Ronsein GE. Comparing Data-Independent Acquisition and Parallel Reaction Monitoring in Their Abilities to Differentiate High-Density Lipoprotein Subclasses. J Proteome Res. 2020;19:248–59. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 [10] Morgantini C, Trifirò S, Tricò D, Meriwether D, Baldi S, Mengozzi A, et al. A short-term increase in dietary cholesterol and fat intake affects high-density lipoprotein composition in healthy subjects. Nutr Metab Cardiovasc Dis. 2018;28:575–81. [11] Plubell DL, Fenton AM, Rosario S, Bergstrom P, Wilmarth PA, Clark WM, et al. High-Density Lipoprotein Carries Markers That Track with Recovery from Stroke. Circ Res. 2020;127:1274–87. [12] Stadler JT, Lackner S, Mörkl S, Trakaki A, Scharnagl H, Borenich A. et al. Obesity Affects HDL Metabolism, Composition and Subclass Distribution. Biomedicines. 2021;9:242. [13] Zhu C, Wong M, Li Q, Sawrey-Kubicek L, Beals E, Rhodes CH, et al. Site- Specific Glycoprofiles of HDL-Associated ApoE are Correlated with HDL Functional Capacity and Unaffected by Short-Term Diet. J Proteome Res. 2019;18:3977. [14] Begue F, Tanaka S, Mouktadi Z, Rondeau P, Veeren B, Diotel N, et al. Altered high-density lipoprotein composition and functions during severe COVID-19. Sci Rep. 2021;11:2291. [15] Annema W, von Eckardstein A. Dysfunctional high-density lipoproteins in coronary heart disease: implications for diagnostics and therapy. Transl Res. 2016;173:30–57. [16] Talbot CPJ, Plat J, Ritsch A, Mensink RP. Determinants of cholesterol efflux capacity in humans. Prog Lipid Res. 2018;69:21–32. [17] Ferraz-Amaro I, Hernández-Hernández MV, Armas-González E, Sánchez- Pérez H, Machado JD, Díaz-González F. HDL cholesterol efflux capacity is 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 related to disease activity in psoriatic arthritis patients. Clin Rheumatol. 2020;39:1871–80. [18] Grao-Cruces E, Millan-Linares MC, Martin-Rubio ME, Toscano R, Barrientos-Trigo S, Bermudez B, et al. Obesity-associated metabolic disturbances reverse the antioxidant and anti-inflammatory properties of highdensity lipoproteins in microglial cells. Biomedicines. 2021;9:1722. [19] Grao-Cruces E, Lopez-Enriquez S, Martin ME, Montserrat-de la Paz S. High-density lipoproteins and immune response: A review. Int J Biol Macromol. 2022;195:117-23. [20] Lim GB. Effect of hypercholesterolaemia on HDL particle remodelling. Nat Rev Cardiol. 2017;14:505. [21] O’Reilly M, Dillon E, Guo W, Finucane O, McMorrow A, Murphy A, et al. High-Density Lipoprotein Proteomic Composition, and not Efflux Capacity, Reflects Differential Modulation of Reverse Cholesterol Transport by Saturated and Monounsaturated Fat Diets. Circulation. 2016;133:1838–50. [22] Padró T, Cubedo J, Camino S, Béjar MT, Ben-Aicha S, Mendieta G, et al. Detrimental Effect of Hypercholesterolemia on High-Density Lipoprotein Particle Remodeling in Pigs. J Am Coll Cardiol. 2017;70:165–78. [23] Andraski AB, Singh SA, Lee LH, Higashi H, Smith N, Zhang B, et al. Effects of Replacing Dietary Monounsaturated Fat With Carbohydrate on HDL (High- Density Lipoprotein) Protein Metabolism and Proteome Composition in Humans. Arterioscler Thromb Vasc Biol. 2019;39:2411–30. [24] Hernáez Á, Sanllorente A, Castañer O, Martínez-González M, Ros E, Pintó X, et al. Increased Consumption of Virgin Olive Oil, Nuts, Legumes, Whole 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Figure 2 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Figure 3 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Figure 4 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Figure 5 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Figure 6 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Figure 7 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Figure 8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 FIGURE LEGENDS Figure 1. Figure A shows the raw data PCA and Figure B the PCA of the dataset after the preprocessing, normalization, and imputation. The Red circles mark samples belonging to subject number 39 and blue circles for subject number 36. Each timepoint is represented by a colored dot: Green for baseline (0 h), Red for postprandial peak (3 h) and Lila for postprandial post-peak. Figure 2. Volcano plot of Basal and Post-peak timepoint, it is graphed Log2 Fold change (B/PP) against -Log10 (p values). Orange points are proteins with a significant differential expression (raw p value < 0.05 and log2FC > |1|); and green points are non-significant proteins. Figure 3. Volcano plot of Peak and Post-peak timepoint, it is graphed Log2 Fold change (P/PP) against -Log10 (p values). Orange points are proteins with a significant differential expression (raw p value < 0.05 and log2FC > |1|); and green points are non-significant proteins. Figure 4. Results from the functional enrichment analysis, each graph represents a database. A, B, and C belong to each database in Gene Ontology (CC stand for Cellular Component, MF for Molecular Function, and BP for Biological Processes), while D are the enriched pathway terms from KEGG and REACTOME. They are ordered firstly by number of genes implicated in each term and then by adjusted p value, only the 20 top terms by this ordering method were shown. Red bars represent terms which more than the half of genes are implicated. Figure 5. Boxplot of each cluster representing the sum of all protein expression profile in each sample. Experimental timepoints are marked by number: Baseline (0 h), postprandial peak (3 h), and postprandial post peak (6 h). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Figure 6. Boxplot of each cluster summarized by timepoint. Experimental timepoints are marked by number: Basal (0 h), postprandial peak (3 h), and postprandial post peak (6 h). Figure 7. REACTOME enrichment analysis for cluster 1, 3, 5, and D (2 + 7). In the X axis is represented the proportion of implicated genes of each cluster that belong to a functional term, red bars mean more than the half of the genes belong to that term. Green box marks the functional term shared between all cluster. Red boxes indicate only the terms shared between cluster 3 and D. Figure 8. Protein-protein interaction network of cluster D, generated by STRING v11.5. This network represents functional and physical proteins association together, proteins used were those belonging to cluster D. The thickness of the edges represents the strength of data support for that association. In general, there are many well-described interactions between the proteins in this cluster. In fact, proteins tend to group in a cluster at the bottom of the image, those proteins are mainly lipoproteins related to lipid metabolism, which is coherent with the functions this cluster modulates. It is not a random network, its protein-protein interaction p value is < 1.0e-16, lower than the confidence level 0.05. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Table 1. Number of proteins per co-expression cluster. Cluster name Number of proteins 1 11 2 9 3 27 4 2 5 50 6 1 7 12 8 7 D (2 + 7) 21 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65