EBioMedicine 43 (2019) 43–53 Contents lists available at ScienceDirect EBioMedicine journal homepage: www.ebiomedicine.com APOA1 oxidation is associated to dysfunctional high-density lipoproteins in human abdominal aortic aneurysm Diego Martínez-López a, Emilio Camafeita b,e, Lídia Cedó c, Raquel Roldan-Montero a, Inmaculada Jorge b,e, Fernando García-Marqués b, María Gómez-Serrano b, Elena Bonzon-Kulichenko b, Francisco Blanco-Vaca c, Luis Miguel Blanco-Colio a,e, Jean-Baptiste Michel d, Joan Carles Escola-Gil c,⁎⁎⁎,1, Jesús Vázquez b,e,⁎⁎,1, Jose Luis Martin-Ventura a,e,⁎,1 a Laboratorio de Patología Vascular, FIIS-Fundación Jiménez Díaz-Universidad Autónoma, Madrid, Spain b Cardiovascular Proteomics Laboratory, Centro Nacional de Investigaciones Cardiovasculares Carlos III (CNIC), Madrid, Spain c Institut d'Investigacions Biomèdiques (IIB) Sant Pau, CIBERDEM, Barcelona, Spain d INSERM U698, Paris, France e CIBER de Enfermedades Cardiovasculares (CIBERCV), Madrid, Spain a r t i c l e i n f o a b s t r a c t Article history: Background: High-density lipoproteins (HDL) are a complex mixture of lipids and proteins with Received 22 December 2018 vasculoprotective properties. However, HDL components could suffer post-translational modifications (PTMs) Received in revised form 15 March 2019 under pathological conditions, leading to dysfunctional HDL. We studied whether HDL are modified in abdominal Accepted 4 April 2019 aortic aneurysm (AAA) and the effect on HDL functionality. Available online 11 April 2019 Methods: HDL were isolated by ultracentrifugation from AAA tissue (HDL-T) and from plasma of healthy volun- teers and then incubated with AAA tissue-conditioned medium (HDL-AAA CM). PTMs from these particles were Keywords: Abdominal aortic aneurysm characterized using Comet-PTM. The ability of HDL-AAA CM for promoting cholesterol efflux was determined HDL ex vivo and in vivo by using J774A.1 [3H]cholesterol-labeled mouse macrophages and after injecting [3H]choles- Oxidative stress terol-labeled mouse macrophages and HDL into the peritoneal cavity of wild-type C57BL/6 mice, respectively. Proteomics Trp50 and Trp108 oxidized forms of APOA1 in HDL incubated with conditioned-medium of activated neutrophils Cholesterol efflux and in plasma of AAA patients and controls were measured by targeted parallel reaction monitoring. Biomarkers Findings: Oxidation was the most prevalent PTM in apolipoproteins, particularly in APOA1. Trp50 and Trp108 in APOA1 were the residues most clearly affected by oxidation in HDL-T and in HDL-AAA CM, when compared to their controls. In addition, cholesterol efflux was decreased in macrophages incubated with HDL-AAA CM in vitro and a decreased macrophage-to-serum reverse cholesterol transport was also observed in mice injected with HDL-AAA CM. Finally, both oxidized Trp50 and Trp108 forms of APOA1 were increased in HDL incubated with conditioned-medium of activated neutrophils and in plasma of AAA patients in relation to controls. Interpretation: Oxidative modifications of HDL present in AAA tissue and plasma were closely associated with the loss of vasculoprotective properties of HDL in AAA. Fund: MINECO, ISCiii-FEDER, CIBERDEM, CIBERCV and LA CAIXA. © 2019 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http:// creativecommons.org/licenses/by-nc-nd/4.0/). 1. Introduction ⁎ Correspondence to: Jose Luis Martin-Ventura, Vascular Research Lab., IIS-Fundación Jiménez Díaz, Autónoma University, Av. Reyes Católicos 2, 28040 Madrid, Spain. ⁎⁎ Correspondence to: Jesús Vázquez, Cardiovascular Proteomics Laboratory, Centro Abdominal aortic aneurysm (AAA) is a permanent focal dilation of Nacional de Investigaciones Cardiovasculares Carlos III (CNIC), Melchor Fernández the abdominal aorta [1]. AAA is characterized by the presence of an Almagro, 3, 28029 Madrid, Spain. Intra Luminal Thrombus (ILT), mainly composed by red blood cells ⁎⁎⁎ Correspondence to: Joan Carles Escola-Gil, Institut d'Investigacions Biomèdiques (IIB), (RBCs), neutrophils and platelets and an aortic wall with few vascular Sant Pau, C/Antoni M Claret 167, 08025 Barcelona, Spain. E-mail addresses: jescola@santpau.cat (J.C. Escola-Gil), jvazquez@cnic.es (J. Vázquez), smooth muscle cells (VSMC) in the remaining media, along with im- jlmartin@fjd.es (J.L. Martin-Ventura). mune cell infiltrates, fibroblasts and neovessels in the adventitia [2]. 1 Equal contribution of last 3 senior authors. Main mechanisms of AAA include proteolysis, oxidative stress and https://doi.org/10.1016/j.ebiom.2019.04.012 2352-3964/© 2019 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). 44 D. Martínez-López et al. / EBioMedicine 43 (2019) 43–53 a protective role of these drugs in preventing AAA growth and rupture Research in context [11–13]. Regardless of the potential contribution of lipids on AAA pro- gression, it is well accepted that low levels of high-density lipoprotein Evidence before this study (HDL) cholesterol are associated to both AAA presence and evolution [14,15]. Moreover, raising HDL inhibits AAA formation and progression Epidemiological studies have clearly demonstrated an inverse as- in experimental models [16,17]. These data support the notion that in- sociation of high-density lipoproteins (HDL) cholesterol levels creasing HDL levels could have a beneficial effect to prevent AAA devel- with different (cardio)vascular diseases (CVD), including abdomi- opment. However, it is nowadays being accepted that the quality of nal aortic aneurysm (AAA). However, different clinical trials HDL is the relevant factor for its vasculoprotective activities [18]. Cardio- aimed at increasing HDL cholesterol levels failed to show benefi- vascular protective functions of HDL include reverse cholesterol trans- cial effects on CVD prevention. In this respect, it has been pro- port (RCT), as well as antioxidant, anti-inflammatory, antiapoptotic posed that HDL cholesterol levels are not causally related with and antithrombotic effects. Regarding AAA, we previously demon- cardiovascular diseases (CVD) and that HDL functionality could strated that HDL particle carries lower alpha-1 antitrypsin and higher underlie the negative association of HDL cholesterol with CVD. In- myeloperoxidase (MPO) proteins, leading to HDL with decreased terestingly, increasing HDL functionality by different approaches antiapoptotic and antioxidant properties [19,20]. Moreover, systemic prevented AAA development in murine experimental models. paraoxonase-1 activity, a surrogate marker of HDL (antioxidant) func- HDL are a complex mixture of lipids and proteins and the tionality, is decreased in AAA patients [21]. More recently, we have vasculoprotective properties of HDL are associated to its compo- shown that AAA patients display impaired HDL cholesterol efflux capac- nents. Previous works including our own have shown a differential ity, the first step of RCT [22]. All these data suggest that HDL could be composition, and a loss of protective functions, of HDL isolated dysfunctional due to impaired concentration or function of specific mo- from AAA patients as compared to HDL from controls. Moreover, lecular components of the particle in AAA. In this respect, APOA1, the it has been demonstrated that HDL and its main component, main protein constituent of HDL, could be oxidatively modified in APOA1, could be oxidatively modified in atherosclerotic plaques, human atherothrombosis, leading to dysfunctional HDL [23,24]. As we leading to dysfunctional HDL. AAA is characterized by increased previously observed that human AAA tissue (ILT and wall) was charac- oxidative responses and by increased tissular APOA1/HDL reten- terized by increased oxidation capacity [20] and by high APOA1 immu- tion, but whether HDL are oxidatively modified in AAA has not nostaining [15] we thus hypothesized that human AAA tissue could be a been explored before. privileged site for searching APOA1/HDL oxidative modifications. In the present paper, we used state of the art proteomics approaches Added value of this study very recently developed [25] to characterize the post-translational modifications (PTMs) on HDL and the changes suffered by the particles By using novel proteomics approaches (COMET-PTM) recently de- in the context of AAA from a completely hypothesis-free perspective. veloped in our lab, we have analyzed the PTMs of HDL recovered We analyzed PTM alterations in HDL isolated from tissue of AAA pa- from AAA tissues. Our results highlight a predominant role of ox- tients. We then studied whether these alterations could be reproduced idative modifications, mainly associated to APOA1, in HDLs. Spe- in plasma-isolated HDL incubated with AAA tissue- conditioned me- cifically, we have shown for the first time a high susceptibility of dium (HDL-AAA CM) and whether they affected HDL functionality. some residues of APOA1 (Trp50 and Trp108) in HDL recovered The ability of HDL-AAA CM for promoting macrophage cholesterol ef- from AAA tissue. Moreover, our results demonstrate that HDL flux was determined ex vivo and in vivo by using mouse macrophages modification under AAA thrombus environment is associated to and after injecting them along with HDL into the peritoneal cavity of dysfunctional HDL, characterized by decreased cholesterol efflux mice. Finally, we explored the potential role of these modifications as capacity in vitro and in vivo. Finally, our results in plasma suggest circulating biomarkers of AAA. that oxidized Trp50 and Trp108 are new potential biomarkers of AAA, potentially reflecting the increased oxidative stress involved 2. Material and methods in this pathology. 2.1. Data statement Implications of all the available evidence Datasets have been deposited in PeptideAtlas (dataset identifier This translational study aimed to investigate potential PTMs of PASS01347). HDL isolated from human tissue AAA samples, supporting an im- portant role of oxidative stress in the mechanisms underlying AAA 2.2. Tissue and tissue-conditioned medium formation. Moreover, we have shown that HDL under AAA envi- ronment loss their main function, namely cholesterol efflux capac- Tissue samples (ILT and wall) were collected during surgical repair ity. Finally, due to the circulating nature of HDL, we have shown from AAA patients and kept at −80° until further processed. Ethical that some of the oxidative forms of APOA1 detected in tissue committee advice and patient informed consent were obtained are also increased in plasma of AAA patients suggesting that (RESAA and AMETHYST studies, CPP Paris-Cochin n° 2095, 1930 and they could be useful as surrogate systemic metrics of tissue-HDL 1931, Inserm Institutional Review Board, IRB0000388). Healthy abdom- functionality. inal aortas were sampled from deceased organ donors with the authori- zation of the French Biomedicine Agency (PFS 09–007). Tissue- conditioned medium was obtained as described [26]. immune-inflammatory response [3]. Improving our understanding of 2.3. HDL isolation AAA pathophysiology could lead to novel diagnostic, prognostic and/ or therapeutic biomarkers. Tissue [100 mg of AAA intraluminal thrombus (HDL-ILT), AAA path- The association of lipids with AAA has been controversial [4–6]. ological wall (HDL-PW) and healthy aortic wall (HDL-HW)] was ho- However, recent genetic studies suggest that lipids could play a role in mogenized and care was taken throughout homogenization to the etiology of AAA [7–10]. Similarly, the effect of statins on AAA maintain a low temperature by keeping the homogenization vessel sub- displayed contradictory results, although more recent studies support merged within liquid N2. After that, lysates were resuspended in buffer D. Martínez-López et al. / EBioMedicine 43 (2019) 43–53 45 (BR20x, FIP, CIP, PMSF, 1 mM EDTA, Tris-HCL pH 8.5). Samples were was performed in accord with the principles outlined in the Declaration vortexed every 15 min for a total of 4 times and kept on ice. of Helsinki, and all participants gave written informed consent. The supernatant was obtained after centrifugation at 12,000 rpm for 15- min. Then, lipoproteins were isolated from these tissue supernatants 2.5. Proteomics or from EDTA plasma samples of healthy volunteers by ultracentrifuga- tion as described [27]. Plasma density was adjusted to ρ = 1.063 g/mL 2.5.1. Sample preparation with KBr and the sample was overlaid with KBr saline solution (ρ = Samples (200 μg of total proteins of HDL) were digested overnight at 1.063 g/mL). Samples were ultracentrifuged at 53000 rpm for 18 h 37 °C in FASP filters with trypsin (Promega, Madison, WI, USA) at an at 4 °C in a TLA-100 rotor (Beckman). After centrifugation, the upper, 40:1 protein:trypsin (w/w) ratio in 50 mM ammonium bicarbonate, apolipoprotein-B-containing fraction was recovered and stored. pH 8.8. The resulting peptides were desalted on C18 Oasis cartridges The bottom fraction, containing whole HDL, was adjusted to 1.25 g/mL (Waters Corporation, Milford, MA, USA) using 50% acetonitrile (ACN) with KBr and overlaid with KBr saline solution (ρ = 1.22 g/mL). (v/v) in 0.1% trifluoroacetic acid (v/v) as eluent, and vacuum dried. The HDL-containing fraction was then ultracentrifuged at 63000 rpm The peptides were then subjected to isobaric TMT labelling following for 24 h at 4 °C. After this step, HDL, in the top layer, were recovered manufacturer's instructions. HDL samples isolated from tissue (healthy as a single band and were extensively rinsed with saline and concen- wall, AAA pathological wall and ILT) were analyzed in three biological trated using a centrifugal concentrating device (cutoff 10 kDa). All replicates. An internal control was constructed by pooling the three HDL samples were desalted by centrifugation and 3 washes with saline. samples from the healthy wall and was used as reference to express rel- In some cases, HDL isolated from plasma were incubated with AAA ative quantification values. HDL isolated from plasma of healthy volun- ILT-conditioned medium (HDL-ILT CM) or with PBS (HDL-control) for teers were incubated with ILT-conditioned media or PBS. An internal 4 h (1:1 proportion), and then reisolated by ultracentrifugation. The control was constructed by pooling HDL incubated with PBS and was protein composition of these particles is shown in Supp. Figs. 1 and 2. used as reference to express relative quantification values. Finally, we isolated neutrophils from blood of healthy volunteers and stimulated 5x106 cells with phorbol 12-myristate 13-acetate (PMA, 2.5.2. LC-MS analysis 3.2uM) (Sigma-Aldrich) for 30 min at 37 °C as described [28]. Then Peptide samples were analyzed using an Easy nano-flow HPLC sys- cells were centrifuged at 12000 rpm 5 min at 4°C and the corresponding tem (Thermo Fisher Scientific, Bremen, Germany) coupled via a supernatant of activated neutrophils was obtained and incubated with nanoelectrospray ion source (Thermo Fisher Scientific) to a Q Exactive HDL for 4 h at 37 °C and then reisolated by ultracentrifugation. HF mass spectrometer (Thermo Fisher Scientific). C18-based reverse phase separation was used with a 2-cm trap column and a 50-cm ana- 2.4. Plasmas lytical column (EASY column, Thermo Fisher Scientific). For the study of PTMs, peptides were loaded in buffer A (0.1% formic Plasmas were obtained from 27 patients with an asymptomatic acid (v/v)) and eluted with a 180-min linear gradient of buffer B (80% infrarenal AAA (aortic size N3 cm). AAA patients were recruited from acetonitrile, 0.1% formic acid (v/v)) at 200 nL/min. Mass spectra were January 2010 to December 2016 at IIS–Fundacion Jimenez Diaz acquired in a data-dependent manner, with an automatic switch be- (Table 1). The presence of cardiovascular risk factors, such as diabetes tween MS and MS/MS using a top 15 method. MS spectra were acquired mellitus, arterial hypertension, dyslipidemia, and smoking habits, in the Orbitrap analyzer with a mass range of 400–1500 m/z and 60,000 were assessed. Patients were considered as diabetics if they were resolution. HCD fragmentation was performed at 27 normalized colli- under treatment (supervised diet, hypoglycemic oral medication, and sion energy and MS/MS spectra were analyzed at 60,000 resolution in insulin) or a basal glycemia N120 mg/dL and/or glycosylated hemoglo- the Orbitrap. bin N6.5%. We defined hypertension as SBP N140 mmHg and/or DBP For the parallel reaction monitoring assays, peptides were loaded in N90 mmHg measured during the examination (after the participant buffer A and eluted with a 90 min linear gradient of buffer B at had been sitting for at least 30 min) or if the participant was already tak- 200 nL/min. MS spectra were acquired in the Orbitrap analyzer with a ing hypotensive medication. Dyslipidemia was diagnosed as the pres- mass range of 400–1500 m/z and 15,000 resolution followed by data- ence of at least 1 of the following characteristics: total cholesterol independent MS/MS spectra of selected precursor ions. HCD fragmenta- N5.2 mmol/L; LDL-cholesterol N3.4 mmol/L, or triglycerides tion was performed at 27 of normalized collision energy and MS/MS N2.3 mmol/L; or the participant was already taking statins or fibrates spectra were analyzed at 60,000 resolution in the Orbitrap. Data analysis medication. Smoking was defined as current smokers or nonsmokers was performed with Xcalibur 2.2 (Thermo Scientific). (including ex-smokers, those who stopped smoking at least 6 months before the inclusion in the study), with a smoker defined by the history 2.5.3. Database search of smoking (≥10 cigarettes per day N1 year). Controls with no AAA (n = For protein identification, searches were performed with Comet re- 28) were recruited from a screening program for AAA between 65-year- lease 2016.01 using 5 ppm precursor ion tolerance, Met oxidation as old men. The absence of AAA was confirmed with a physical examina- a dynamic modification, trypsin digestion with 1 missed cleavage tion and an ultrasound scan (aortic size b3 cm). (unless otherwise specified) and fixed Cys carbamidomethylation The ethical committee on human research at IIS, Fundacion Jimenez (57.021464 Da) [29,30]. TMT labeling at N-terminal end and Lys was Dıaz, Autonoma University (Madrid, Spain) approved the study, which also considered as a fixed modification (229.162932 Da). Fragment ion tolerance was 0.02 bin, 0 mass offset. Precursor tolerance type and iso- tope error were set to 1. Precursor charge range was 2–4, maximum precursor charge 5 and maximum fragment charge 3. Only y- and b- Table 1 ions were used for scoring. Peptide identification from MS/MS data Clinical characteristics of controls and AAA patients. was performed using the probability ratio method [31]. False discovery Control (n = 28) AAA patients (n = 27) p-Value rates (FDR) of peptide identifications were calculated using the refined method [32]; 1% FDR was used as the default criterion for peptide Age (years) 64.9 ± 0.1 67.7 ± 0.9 b0.001 Sex (male/female) 28/0 27/0 identification. Hypertensive (%) 53 ± 9 63 ± 9 0.221 Smokers (%) 64 ± 9 40 ± 9 0.463 2.5.4. Comet-PTM Diabetes (%) 32 ± 8 25 ± 8 0.322 For PTM analysis, open searches using a precursor ion tolerance of Dyslipidemic (%) 50 ± 9 44 ± 9 0.564 500 Da were performed using Comet-PTM, as described [25]. Comet- 46 D. Martínez-López et al. / EBioMedicine 43 (2019) 43–53 D. Martínez-López et al. / EBioMedicine 43 (2019) 43–53 47 PTM is a novel open search algorithm that calculates the difference be- monitoring of conjugated diene formation at 37 °C for 1 h. The lag tween theoretical and experimental precursor mass (DMass), and adds phase was calculated from the intersection point between the maximal up this mass iteratively to each one of the amino acid masses in the pep- slope of the curve and initial absorbance. tide sequence, calculating a Xcorr score in each one of the possible mod- ified forms of the peptide. The selected candidate is the modified 2.8. Cholesterol efflux analyses peptide form that produces the highest Xcorr. Comet-PTM produces the same score than Comet with a preselected modification, with the Cholesterol efflux capacity to HDL-control or HDL-ILT CM (50 μg/mL difference that Comet-PTM identifies all possible modifications within APOA1) was determined by using J774A.1 [3H]-cholesterol-labeled + − 500 Da of precursor ion [25]. The accuracy of Comet-PTM to assign macrophages after loaded with acetylated (ac)-LDL as previously re- the modification to the correct site has been estimated to be between 80 ported [36]. Ten to 12 weeks-old female wild-type C57BL/6 mice re- and 85% using both high-throughput data and synthetic peptides [25]. ceived intraperitoneal injections of HDL-control or HDL-ILT CM (100 μg APOA1) together with [3H]cholesterol-labeled J774 macrophages. 2.5.5. SHIFTS Mice were randomized into two groups and the HDL injections were SHIFTS (Systematic Hypothesis-free Identification of modifications performed blinded by the technician. Cell viability of the macrophage with controlled FDR based on ultra-Tolerant database Search) was preparations ranged from 85 to 90% and each mouse received an i.p. used to identify peaks in the DMass distribution, assign PSMs to peaks dose of [3H]cholesterol-labeled mouse macrophages (0.92 × 105 cells/ and calculate the FDR for peptide identification. Further details about mouse, 0.99 × 105 cpm/mouse). Serum [3H]cholesterol was determined SHIFTS are described in.25 after 3 h of the macrophage injection as described [36]. All mice were kept in a temperature-controlled environment (20 °C) with a 12-h 2.5.6. Annotation of modifications light/dark cycle and maintained on a regular chow diet (Safe, Scientific An in-house developed script was used for the semisupervised an- Animal Food & Engineering; A04C, 3% fat, 2900 kcal/kg). Food and water notation of peptide modifications as described [25]. Briefly, the script were provided ad libitum. All animal procedures were conducted in ac- matches DMass values against the Unimod database, considering the cordance with published regulations and reviewed and approved by the amino acid modification positions revealed by Comet-PTM as well as Institutional Animal Care Committee of the Institut de Recerca of the the preceding and following residues. For simplicity, artifactual modifi- Hospital de la Santa Creu i Sant Pau (ref. 7281). cations introduced during sample preparation and modifications that could not be annotated in Unimod were not considered in further anal- 2.9. Statistics ysis. Met mono-oxidation (+16 Da) is not considered as a PTM. Results are shown as mean ± standard deviation. In vitro and in vivo 2.5.7. Peptide quantification and statistical analysis data was analyzed by t-test or Mann-Whitney test, respectively. Multi- The quantitative information from TMT reporter intensities was in- variate logistic regresion analysis was completed to identify the impact tegrated from the spectrum level to the peptide level and then to the of different variables on the prediction of AAA. Statistical differences in protein level on the basis of the WSPP model using the Generic Integra- c-statistics were compared using the method by DeLong et al. [37]; tion Algorithm (GIA), as described [25,33,34]. The algorithm was modi- 95% CIs were calculated for each comparison. P value b.05 was consid- fied to incorporate the quantitative values of modified peptides as part ered statistically significant. of the automated workflow [25]. Hence, quantitative peptide values are referred to the weighted averages of the non-modified peptides 3. Results from the same protein and therefore are not affected by protein abun- dance changes [25]. This statistical model describes accurately the 3.1. Characterization of PTMs of HDL-like particles recovered from AAA tis- error distribution of abundance changes of both non-modified and sue highlights APOA1 oxidation modified peptides in null hypothesis experiments [25]. Significant abundance changes of modified peptides across the different samples To analyze whether HDL are modified in the context of AAA, we de- were detected by Student's t-test. cided to recover HDL-like particles from homogeneized AAA tissue [ILT (HDL-ILT) and pathological wall (HDL-PW)] and from healthy aortic 2.6. Immunohistochemistry wall (HDL-HW) by ultracentrifugation. High-throughput analysis of PTMs using Comet-PTM revealed that Samples of ILT and wall obtained from AAA and healthy arterial wall one third of peptides from lipid metabolism proteins had modifications were embedded in paraffin, and 4 μm cross-sections were cut. Immuno- that could be assigned to known PTMs, while two-thirds of the peptides histochemistry was performed with antibodies against APOA1 (ab7613, were non-modified (Fig. 1A). Oxidation accounted for the majority of abcam) and the lipid peroxidation marker MDA (ab6463, abcam), as de- modifications in HDL-like particles (Fig. 1A and Supp. Table 1). The ma- scribed [35]. jority of PTMs were found in APOA1 (Fig. 1B). Of note, the proportion of modified peptides in APOA1 was higher than in the rest of proteins 2.7. Lipoprotein oxidation (Fig. 1C). Oxidative modifications in APOA1 affected mainly Trp resi- dues, while Tyr, Pro, His, Asp, Met and Lys were also affected, although We measured basal HDL conjugated dienes at 234 nm by using a 96- to a lesser extent (Fig. 1D). Quantitative analysis taking as reference microwell plate for UV detection. HDL was also oxidized with 2.5 μM the peptides from the same protein that were not modified revealed CuSO4 and the oxidation kinetics was followed by continuous that the pattern of PTM changes suffered by APOA1 peptides was Fig. 1. PTMs in lipid metabolism proteins in HDL recovered from AAA tissue. (A) Peptide-spectrum matches (PSM) distribution of posttranslationally modified and unmodified peptides among proteins pertaining to lipid metabolism. (B) Adscription of the PSMs corresponding to posttranslationally modified peptides to lipid metabolism proteins. (C) PSM distribution of posttranslationally modified and unmodified peptides from APOA1. (D) PSM distribution of oxidative PTMs in APOA1 according to the modified residue and their abundance change (yellow, increased in thrombus; violet, decreased). (E) Abundance changes of oxidatively modified ApoA1 peptides as compared to healthy tissue expressed in terms of the standardized log2 ratio Zpq [n = 3, p b .05 labeled in green, t-test]; other differentially abundant modifications are shown in Supp. Fig. 4. (F) Colocalization of ApoA1 and malondialdehyde in human AAA tissue. No staining for APOA1 or malondialdehyde is observed in negative controls (scale bar = 100uM). DeA: Deamidation; MethylMal: Methylmalonylation;Phos: Phos- phorylation; Methyl: Methylation; AGE: Advanced glycation end product; Kyn: Kynurenine; ox, 2ox and 3ox: addition of one, two or three oxygens, respectively. Peptides with a single oxidation in Met (marked in light blue in A and C) are included in the group of unmodified peptides. 48 D. Martínez-López et al. / EBioMedicine 43 (2019) 43–53 D. Martínez-López et al. / EBioMedicine 43 (2019) 43–53 49 remarkably similar in HDL-ILT and HDL-PW in comparison with those of HDL-HW (Fig. 1E). A generalized increase in the abundance of peptides bearing oxidized Trp forms in HDL-ILT and HDL-PW in comparison with HDL-HW was observed (Fig. 1D). The peptides contained mono-, di- and tri-oxidations and also Trp in the form of kynurenine (Fig. 1E). Pep- tide forms containing the four Trp residues of APOA1 were detected by mass spectrometry; from these, oxidation was found in three of them, corresponding to Trp50, Trp72 and Trp108. This trend was not detected in peptides bearing oxidized forms in other aminoacids (Fig. 1E). Inter- estingly, we found by immunohistochemistry that APOA1 in ILT and wall of AAA colocalizes with malonaldehyde (MDA), a marker of lipid peroxidation, which was not present in healthy aortic wall (Fig. 1F). 3.2. Incubation of plasmatic HDL with AAA tissue-conditioned medium pro- Fig. 3. Macrophage cholesterol efflux to HDL-ILT CM and HDL-control in vitro and duces APOA1 oxidation and impacts HDL functionality macrophage -to-serum reverse cholesterol transport (RCT) in vivo. (A) Ability of HDL- ILT CM and HDL-control (HDL-C) for promoting cholesterol efflux from J774A.1 [3H] cholesterol-labeled mouse macrophages. After 4 h of incubation, efflux of cholesterol Our previous results indicated that HDL-like particles retained in was determined and expressed as [3H]cholesterol medium /([3H]cholesterol cells + AAA tissue are affected by specific oxidative modifications. We hypoth- [3H]cholesterol medium) x 100. Values are means ± SEM of four independent HDL per esized that oxidizing agents present in the AAA tissue caused this oxida- group. (B) Macrophage-to-serum RCT after injecting intraperitoneally with [3H] tive damage on HDL and could affect HDL functionality. cholesterol-labeled mouse macrophages and HDL-ILT CM or HDL-control. Macrophage- In an attempt to reproduce the modifications taking place in the AAA derived [3H]cholesterol radioactivity was measured at 3 h post-macrophage injection in serum. Values are means ± SEM (N = 8 individual mice for HDL-control, N = 7 for tissue environment, we performed an ex vivo approach preparing HDL HDL-ILT CM). [*p ≤ .05 vs HDL-control, t-test]. isolated from plasma of healthy volunteers and incubating them with tissue-conditioned medium (CM) during 4 h; then, HDL were reisolated by ultracentrifugation and subjected to proteomics analysis following was reduced in the HDL-ILT CM injected group as compared to HDL- the same approach as above. However, as the limited amount of healthy control (Fig. 3B). wall (and then, healthy-CM) to be used in the functional in vivo studies made this task unfeasible, we decided to use HDL incubated with PBS 3.3. Increased oxidized Trp50 and Trp108 of APOA1 in HDL incubated with (HDL-control). We firstly confirmed that healthy wall-CM did not pro- PMN-conditioned medium and in plasma of patients with AAA duce oxidative HDL modifications when compared to HDL-control (Supp. Fig. 3A). Analysis of HDL incubated with ILT-conditioned me- Given the increased abundance of oxidized Trp in APOA1 observed dium (HDL-ILT CM) showed that the basal proportions of PTMs were in HDL from AAA tissue or in HDL incubated with AAA-tissue condi- similar to those of HDL-like particles from ILT, being oxidative modifica- tioned media, we tested potential cellular sources of ROS involved in tions the most abundant and APOA1 the protein mostly affected by PTM these oxidative processes. As it has been previously shown that MPO (Fig. 2 A-C and Supp. Table 1). Quantitative analysis confirmed the gen- is involved in APOA1 oxidation [23,24] and we previously showed in- eralized increase of peptides bearing oxidized Trp forms in the HDL-ILT creased MPO in HDL and PMN of AAA patients [19,28], we hypothesized CM samples in comparison to those from HDL-control (Fig. 2D). In this that oxidants (including MPO) released by activated PMNs could medi- case, oxidation affected to Trp50 and Trp108 (Fig. 2E). Finally, to analyze ate the oxidation of APOA1 in AAA. Then, we incubated HDL isolated if incubation of HDL with ILT CM also produces oxidation at the lipid from plasma of healthy volunteers with the conditioned medium ob- level, we evaluated the basal HDL conjugated dienes. HDL-ILT CM had tained after activation of neutrophils with PMA; after that, HDL were significantly increased basal levels of conjugated dienes when com- reisolated by ultracentrifugation. pared with those of HDL-control (Fig. 2F). Moreover, we also evaluated To test specific APOA1 oxidation, we setup a targeted mass spec- the potential of HDL-ILT CM to be further oxidized in an experimental trometry assay (Fig. 4A and B) and used it to quantify non-modified system, in which isolated HDL were exposed to the prooxidant Cu2+. and oxidized Trp (Trp8, 50, 72 and 108)-containing peptides in HDL in- Under these conditions, HDL-ILT CM exhibited a significantly shorter cubated with conditioned medium of PMA-activated PMNs. The oxi- lag phase when compared with HDL-control, indicating that extensive dized form of Trp8 was not detected in this preparation. We observed oxidation was promoted earlier in the HDL-ILT CM (Fig. 2G). an increase of oxidation forms of Trp 50 and Trp 108 in HDL- Then, we used the HDL-ILT CM particles to interrogate the functional incubated with PMN-conditioned media (Fig. 4C). Moreover, oxidized consequences that oxidative modifications may produce on HDL. We Trp72 was also increased (not shown), as previously observed in ath- firstly confirmed that incubation of HDL with healthy wall-CM did not erosclerotic plaques [38] and here, in AAA tissue. affect the ability of macrophages to act as acceptors of cholesterol Finally, we hypothesized that oxidized Trp in APOA1 found in tissue ex vivo when compared with HDL-control (Supp. Fig. 3B). In contrast, could also be present in plasma, where they could be used as circulating when HDL-ILT CM particles were added to macrophages, we found surrogate markers of HDL dysfunction. To explore this possibility, we that their ability to promote cholesterol efflux was blunted compared used the same targeted mass spectrometry assay described above. The with HDL-control (Fig. 3A). Moreover, to study the in vivo effect of oxidized forms of Trp8 and Trp72 could not be directly detected in HDL-ILT CM, [3H]cholesterol-labeled J774A.1 macrophages were plasma; however, we could accurately monitor the oxidized Trp50 injected together with HDL into the peritoneal cavity of mice. Three and Trp108 peptide forms in control subjects and AAA patients. The pro- hours after injection of the macrophages, [3H]cholesterol in plasma portions of the dioxidized forms of both Trp50 and Trp108 were Fig. 2. PTMs in lipid metabolism proteins from HDL incubated with AAA ILT-conditioned medium. (A, B, C, D and E) These graphs have the same meaning than in Fig. 1 [n = 4, p b .05, labeled in green, t-test]. Other differentially abundant modifications are shown in Supp. Fig. 5. (F) Basal levels of conjugated dienes in HDL-ILT CM compared to those found in HDL-control (HDL-C). (G) Lag phases of HDL-ILT CM and HDL-C in the presence of CuSO4. A shorter lag phase indicates a higher susceptibility to be extensively oxidized under prooxidant conditions. Values are means ± SEM of four independent HDL per group. *p ≤ .05 vs HDL-C [t-test]. For simplicity, in (A) and (B) methionine-oxidized peptides (marked in light blue) are included in the group of unmodified peptides. DeA: Deamidation, MethylMal: Methylmalonylation, Phos: Phosphorylation, Methyl: Methylation. 50 D. Martínez-López et al. / EBioMedicine 43 (2019) 43–53 D. Martínez-López et al. / EBioMedicine 43 (2019) 43–53 51 significantly increased in plasma of AAA patients as compared to con- Table 2 trols (Fig. 4C). Multivariate statistical analysis of Trp50 and Trp108. Multivariate logistic regression analysis also revealed that the pro- A) Plasma levels of dioxidized Trp50 and Trp108 in ApoA1 are independent portions of each one of the two dioxidized Trp forms were significantly predictors of AAA associated to the presence of AAA after adjustment by age (Table 2A). p-Value OR (95% CI) (1) ROC analysis revealed that the proportions of the two dioxidized Trp W50diox 0.008 2.22 (1.23–4.01) forms of APOA1 were good predictors of the disease (Table 2B). More- W108diox 0.013 2.52 (1.21–5.25) over, a panel with the two dioxidized Trp forms significantly increased (1) OR for variables expressed in units of standard deviation. Logistic regression the power to predict AAA over age alone (Table 2B). was adjusted by age; no other risk factors were found to associate with AAA B) Plasma levels of dioxidized Trp50 and Trp108 in ApoA1 improve prediction of 4. Discussion AAA over risk factors AUC (95% CI)1 p-Value2 p-Value3 In this work we have used state of the art proteomics approaches to characterize the whole landscape of PTM in HDL and the specific alter- Age 0.65 (0.48–0.82) 0.050 W50diox 0.70 (0.56–0.84) 0.010 ations under AAA tissue environment. We have used as models HDL- W108diox 0.79 (0.67–0.91) 0.020 like particles recovered from AAA tissue and HDL isolated from plasma Age, W50diox and W108diox 0.84 (0.73–0.95) b0.001 0.01 incubated with AAA tissue-conditioned media. To our knowledge, this 1 Area under the curve obtained by ROC analysis. is the first time that such a comprehensive, quantitative and 2 Statistical significance vs reference value (AUC = 0.5). hypothesis-free analysis of PTM has been performed in HDL particles. 3 Statistical significance vs Age, DeLong's test. It is interesting to note that the majority of modifications found in HDL are of oxidative nature; although HDL are known to suffer oxidative modifications, the proportion of these modifications with respect to that oxidative modifications similar to those observed in HDL-like particles of unmodified peptides and other potential modifications has never from AAA tissue and HDL-ILT CM. Among those oxidative modifications, been studied before. Our analysis revealed that the major hallmark of it has been previously observed that MPO is able to oxidize Trp72 [38], AAA is the oxidative modifications on Trp residues in APOA1. We which was confirmed in our in vitro study. These data suggest that were able to detect all Trp-containing peptides in APOA1, and those af- neutrophil-derived oxidants may play a relevant role in APOA1/HDL ox- fected by oxidation were found in different oxidation states, including idation in AAA tissue. We cannot, however, discard the contribution of not only mono- and di-oxidation, but also in kynurenine form, reflecting other sources of ROS such as red blood cell-derived ferrous iron (highly gradual stages of oxidative damage. Other residues, like Tyr and Pro, abundant in AAA) [39] or other enzymatic sources. In these regard, neu- were also affected but to a lesser extent. Pro residues were presumably trophil extracellular trap-derived enzymes such as nicotinamide ade- hydroxylated. A remarkable finding of this study is that the abundance nine dinucleotide phosphate (NADPH) oxidases have been shown to of oxidative modifications in Trp was consistently increased in HDL- oxidize HDL in autoinmune diseases [40]. like particles obtained from AAA tissue, a trend that was not detected The best-known HDL's protective function is their ability to promote in other residues. This apparently selective effect of AAA-derived oxi- cholesterol efflux from cells, including that from macrophages in the ar- dants on Trp residues was confirmed when HDL particles isolated terial wall. In the present study, we show that HDL-ILT CM particles are from plasma of healthy individuals were incubated with conditioned characterized by decreased cholesterol efflux capacity (CEC). Since Trp media obtained from ILT. All these results suggested that Trp residues residues are key factors regulating the MPO-mediated loss of the choles- in APOA1 are the most sensitive to oxidants present in the human terol acceptor activity of APOA1 [41], the oxidative modifications in Trp AAA environment. These oxidants were also manifested by the presence residues described in this study could partially explain the decreased of lipid oxidation products whose tissular distribution closely CEC observed under our experimental conditions, although we cannot colocalized with that of APOA1, suggesting the presence of oxidized discard the contribution of other modifications. To further assess this ef- lipids and proteins in HDL particles retained in the AAA tissue. fect in vivo, we used the peritoneal cavity as a surrogate to evaluate the Previous papers have shown that the oxidative modification of dif- potential impact of HDL-ILT CM on macrophage-to-serum RCT in vivo. ferent residues of APOA1 are present in atherosclerotic plaques and Because of the highly turnover of HDL particles in the peritoneal fluid are associated to HDL dysfunctionality [23,24]. In this work, by combin- and the highly dynamic nature of the cholesterol efflux assay, we short- ing immunohistochemical and biochemical approaches in addition to ened our macrophage-specific RCT assay for directly assessing the local proteomics, we demonstrate for the first time that APOA1/HDL is oxi- effects of HDL-ILT CM, showing a decreased macrophage-to-serum dized within the AAA tissue environment. ILT in human AAA is charac- transport in mice injected with HDL-ILT CM. The importance of effective terized by the presence of neutrophils and RBC, which are potent cholesterol efflux in AAA is due to the fact that cholesterol crystals are sources of oxidative stress through the release of MPO and its present in AAA wall, where they can contribute not only to immune re- potentialization by redox-active Fe++ from free hemoglobin. We previ- sponses but also to SMC death, two main mechanisms underlying AAA ously demonstrated that the levels of MPO are higher in human AAA tis- progression. Whether therapies aimed at increasing HDL cholesterol ef- sue and tissue-conditioned medium [16]. MPO mainly released by flux could prevent AAA progression deserves further studies. Interest- activated neutrophils, catalyzes the reaction between hydrogen perox- ingly, we previously demonstrated that increasing HDL function by ide and chloride ions to produce hypoclorous acid as the primary oxi- using 4DF peptide (which is resistant to APOA1 oxidation) or by over- dant. Interestingly, MPO is associated to HDL and we showed that HDL expressing the major HDL-antioxidant protein (PON-1), was able to de- from patients with AAA carries higher MPO levels as compared to HDL crease experimental AAA formation [15,21]. from controls [19]. Specifically, we have now demonstrated that HDL in- There is a great interest to find biomarkers of HDL quality and func- cubated with conditioned medium of activated neutrophils display tion that can improve risk prediction and target drug treatments. Efforts Fig. 4. Targeted quantitative assay of Trp oxidation in HDL incubated with activated neutrophils and plasma from AAA patients. (A) Quantitative PRM analysis of Trp-containing peptides in plasma. The intensities of the y7+ and y12+ fragments were used to quantify the different oxidation states of Trp108 (left panels) and Trp50 (right panels), respectively. (B) Extracted ion chromatograms corresponding to these fragments for representative AAA and control plasma samples. (C) The abundance of oxidized Trp50 and Trp108 was increased in HDL incubated with the supernatant (SN) of PMA-activated neutrophils as compared to HDL incubated with PBS [4 h, n = 3–6, ** p b .01 and * p b .05, t-test]. (D)The concentrations of the two dioxidized Trp50 and Trp108 forms are significantly increased in AAA patients (N = 27) in relation to controls (N = 28) [* p b .05; *** p b .01, Mann-whitney test]. The abundance of the dioxidized Trp forms is expressed as percentage in relation to the non-modified peptide. 52 D. Martínez-López et al. / EBioMedicine 43 (2019) 43–53 to develop reproducible, cost-effective, validated assays that measure (PI1600139), and FEDER “Una manera de hacer Europa”, CIBERDEM the potential functions of HDL are recognized as a major challenge in and CIBERCV are Instituto de Salud Carlos III projects. The project lead- the cardiovascular field. Growing evidence indicates an inverse relation- ing to these results has also received funding from “la Caixa” Banking ship between HDL CEC and the incidence of CVD [42]. We have recently Foundation under the project code HR17-00247. The CNIC is supported shown an impairment of macrophage CEC in ApoB depleted plasma by the MINECO and the Pro-CNIC Foundation, and is a Severo Ochoa ("plasma HDL") from AAA patients [22]. In addition to the analysis of Center of Excellence (MINECO award SEV-2015-0505). CEC in plasma HDL, other authors have measured oxidized forms of cir- culating APOA1 in CVD [38,43]. In general, the authors focused on oxi- Declaration of interest dized forms of APOA1 which negatively regulated HDL functions and that were detected in atherosclerotic plaques. In the largest study per- All authors have reported that they have no relationships relevant to formed to date, oxidized APOA1 (Trp72) was assessed in plasma using the contents of this paper to disclose. specific antibodies [38]. In this work, we observed that oxidized forms of Trp in APOA1 were increased within AAA tissue environment, and Author contributions also in HDL incubated with activated neutrophils, and tested whether these forms could serve to monitor a pathophysiologically relevant pro- Diego Martínez-López = HDL isolation, TMT studies, PTM and PRM cess occurring locally during AAA. Although we detected peptides con- analysis, revising the work and final approval of the manuscript. taining Trp72 by targeted mass spectrometry, we were unable to detect Emilio Camafeita = PRM analysis, revising the work and final ap- their oxidized forms in plasma. However, we could readily detect the proval of the manuscript. oxidized forms of Trp50 and Trp108, finding that their relative propor- Lídia Cedó = Functional experiments (Efflux and RCT), revising the tions are increased in plasma of AAA patients when compared to con- work and final approval of the manuscript. trols. It is also interesting to note that the oxidized forms of Trp50 and Raquel Roldan-Montero = HDL isolation and functional experi- Trp108 were independently associated to AAA. Moreover, the model in- ments,revising the work and final approval of the manuscript. cluding the oxidized forms of both Trp50 and Trp108 significantly im- Inmaculada Jorge, Fernando García-Marqués, María Gómez-Serrano, proved the accuracy of AAA prediction over that provided by known Elena Bonzon-Kulichenko = TMT studies and COMET-PTM analysis,re- risk factors alone. However, for a potential translation to the clinics, vising the work and final approval of the manuscript. these data must be validated in independent cohorts and novel ap- Francisco Blanco-Vaca, Luis Miguel Blanco-Colio = Interpretation of proaches based on antibodies specifically recognizing the modified pep- the data, revising the work and final approval of the manuscript. tides should be developed. On the whole, our in vitro and in vivo data Jean-Baptiste Michel = Providing human AAA biobank, revising the support the association of these novel oxidized forms of APOA1 with work and final approval of the manuscript. dysfunctional HDL in AAA patients. Joan Carles Escola-Gil, Jesús Vázquez, Jose Luis Martin-Ventura = Conception and design of the study, analysis of results and writing the 4.1. Limitations manuscript. One main limitation of our study is the scarce amount of APOA1 in Appendix A. Supplementary data HDL-like particles that can be obtained from AAA tissue, which makes it unfeasible to perform functional in vivo studies. This low level is Supplementary data to this article can be found online at https://doi. probably also potentially related to APOA1 oxidation, that promotes org/10.1016/j.ebiom.2019.04.012. APOA1 dissociation [38]. To circumvent this problem, we have used HDL isolated from plasma of healthy volunteers incubated ex vivo References with ILT-conditioned media, which contains a representative patho- physiological composition of AAA molecules. Moreover, it could have [1] Sakalihasan N, Limet R, Defawe OD. Abdominal aortic aneurysm. Lancet 2005;365: 1577–89. been desirable to compare the effects of ILT-CM with that of healthy- [2] Michel J-B, Martin-Ventura J-L, Egido J, Sakalihasan N, Treska V, Lindholt J, et al. CM instead of HDL incubated with PBS (HDL-control). However, no ox- Novel aspects of the pathogenesis of aneurysms of the abdominal aorta in humans. idative or functional differences were observed between HDL incu- Cardiovasc Res 2011;90:18–27. [3] Torres-Fonseca M, Galan M, Martinez-Lopez D, Cañes L, Roldan-Montero R, Alonso J, bated with healthy-CM and HDL-control. Moreover, the oxidation of et al. Pathophisiology of abdominal aortic aneurysm: biomarkers and novel thera- APOA1 in HDL-like particles recovered from AAA tissues preferentially peutic targets. Clin Investig Arterioscler 2018. https://doi.org/10.1016/j.arteri.2018. affects the same residues (Trp50 and Trp108) when using the ex vivo 10.002 pii: S0214-9168(18)30139-6. [Epub ahead of print]. model of HDL-ILT CM. Thus, despite the potential limitations of recov- [4] Forsdahl SH, Singh K, Solberg S, Jacobsen BK. Risk factors for abdominal aortic aneu- rysms: a 7-year prospective study: the Tromsø study, 1994–2001. Circulation 2009; ering APOA1/HDL from tissue or obtaining healthy-CM to perform 119:2202–8. functional experiments, the present study showed that the methods [5] Louwrens HD, Adamson J, Powell JT, Greenhalgh RM. Risk factors for atherosclerosis used herein to examine HDL in the AAA environment could be relevant in men with stenosing or aneurysmal disease of the abdominal aorta. Int Angiol 1993;12:21–4. for testing the quantity and functional significance of modified HDL [6] Rizzo M, Krayenbühl P-A, Pernice V, Frasheri A, Battista Rini G, Berneis K. LDL size in vivo. and subclasses in patients with abdominal aortic aneurysm. Int J Cardiol 2009; On the whole, our results indicate that APOA1 oxidation in AAA 134:406–8. [7] Weng L-C, Roetker NS, Lutsey PL, Alonso A, Guan W, Pankow JS, et al. Evaluation of patients can contribute to dysfunction of HDL particles in AAA charac- the relationship between plasma lipids and abdominal aortic aneurysm: a Mende- terized by decreased cholesterol efflux capacity. Moreover, our prelimi- lian randomization study. PLoS One 2018;13:e0195719. nary data in plasma suggest new potential biomarkers (oxidized Trp50 [8] Harrison SC, Holmes MV, Burgess S, Asselbergs FW, Jones GT, Baas AF, et al. Genetic Association of Lipids and Lipid Drug Targets with Abdominal Aortic Aneurysm: a and Trp108 on APOA1) that could be useful as a surrogate systemic met- meta-analysis. JAMA Cardiol 2018;3:26–33. rics of tissue-HDL functionality. [9] Klarin D, Damrauer SM, Cho K, Sun YV, Teslovich TM, Honerlaw J, et al. Genetics of blood lipids among ~300,000 multi-ethnic participants of the million veteran pro- gram. Nat Genet 2018;50:1514–23. Funding [10] Jones GT, Tromp G, Kuivaniemi H, Gretarsdottir S, Baas AF, Giusti B, et al. Meta- analysis of genome-wide association studies for abdominal aortic aneurysm iden- This work was partly funded by the Spanish MINECO (SAF2016- tifies four new disease-specific risk loci. Circ Res 2017;120:341–53. 80843-R and BIO2015-67580-P), Fondo de Investigaciones Sanitarias [11] Twine CP, Williams IM. Systematic review and meta-analysis of the effects of statin therapy on abdominal aortic aneurysms. Br J Surg 2011;98:346–53. ISCiii-FEDER (Biobancos RD09/0076/00101, and ProteoRed PRB2 [12] Cheng W, Jia X, Li J, Cheng W, Liu Z, Lin Z, et al. Relationships of statins therapy and IPT13/0001 and PRB3 PT17/0019), Instituto de Salud Carlos III Hyperlipidemia with the incidence, rupture, post-repair mortality and all-cause D. Martínez-López et al. / EBioMedicine 43 (2019) 43–53 53 mortality of abdominal aortic aneurysm and cerebral aneurysm: a meta-analysis [28] Ramos-Mozo P, Madrigal-Matute J, Martinez-Pinna R, Blanco-Colio LM, Lopez JA, and systematic review. J Cardiovasc Pharmacol 2019;73:232–40 [Jan 25]. Camafeita E, et al. Proteomic analysis of polymorphonuclear neutrophils identifies [13] Salata K, Syed M, Hussain MA, de Mestral C, Greco E, Mamdani M, et al. Statins re- catalase as a novel biomarker of abdominal aortic aneurysm: potential implication duce abdominal aortic aneurysm growth, rupture, and perioperative mortality: a of oxidative stress in abdominal aortic aneurysm progression. Arterioscler Thromb systematic review and meta-analysis. J Am Heart Assoc 2018;7:e008657. Vasc Biol 2011;31:3011–9. [14] Golledge J, van Bockxmeer F, Jamrozik K, McCann M, Norman PE. Association be- [29] Eng JK, Jahan TA, Hoopmann MR. Comet: an open-source MS/MS sequence database tween serum lipoproteins and abdominal aortic aneurysm. Am J Cardiol 2010;105: search tool. Proteomics 2013;13:22–4. 1480–4. [30] Eng JK, Hoopmann MR, Jahan TA, Egertson JD, Noble WS, MacCoss MJ. A deeper look [15] Burillo E, Lindholt JS, Molina-Sánchez P, Jorge I, Martinez-Pinna R, Blanco-Colio LM, into Comet–implementation and features. J Am Soc Mass Spectrom 2015;26: et al. ApoA-I/HDL-C levels are inversely associated with abdominal aortic aneurysm 1865–74. progression. Thromb Haemost 2015;113:1335–46. [31] Martínez-Bartolomé S, Navarro P, Martín-Maroto F, López-Ferrer D, Ramos- [16] Krishna SM, Seto SW, Moxon JV, Rush C, Walker PJ, Norman PE, et al. Fenofibrate in- Fernández A, Villar M, et al. Properties of average score distributions of SEQUEST: creases high-density lipoprotein and sphingosine 1 phosphate concentrations limit- the probability ratio method. Mol Cell Proteomics 2008;7:1135–45. ing abdominal aortic aneurysm progression in a mouse model. Am J Pathol 2012; [32] Navarro P, Vázquez J. A refined method to calculate false discovery rates for peptide 181:706–18. identification using decoy databases. J Proteome Res 2009;8:1792–6. [17] Torsney E, Pirianov G, Charolidi N, Shoreim A, Gaze D, Petrova S, et al. Elevation of [33] Navarro P, Trevisan-Herraz M, Bonzon-Kulichenko E, Núñez E, Martínez-Acedo P, plasma high-density lipoproteins inhibits development of experimental abdominal Pérez-Hernández D, et al. General statistical framework for quantitative proteomics aortic aneurysms. Arterioscler Thromb Vasc Biol 2012;32:2678–86. by stable isotope labeling. J Proteome Res 2014;13:1234–47. [18] Favari E, Thomas MJ, Sorci-Thomas MG. High-density lipoprotein functionality as a [34] García-Marqués F, Trevisan-Herraz M, Martínez-Martínez S, Camafeita E, Jorge I, new pharmacological target on cardiovascular disease: unifying mechanism that ex- Lopez JA, et al. A novel systems-biology algorithm for the analysis of coordinated plains high-density lipoprotein protection toward the progression of atherosclerosis. protein responses using quantitative proteomics. Mol Cell Proteomics 2016;15: J Cardiovasc Pharmacol 2018;71:325–31. 1740–60. [19] Ortiz-Muñoz G, Houard X, Martín-Ventura J-L, Ishida BY, Loyau S, Rossignol P, et al. [35] Fernandez-García CE, Tarin C, Roldan-Montero R, Martinez-Lopez D, Torres Fonseca HDL antielastase activity prevents smooth muscle cell anoikis, a potential new M, Lindhot JS, et al. Increased galectin-3 levels are associated with abdominal aortic antiatherogenic property. FASEB J 2009;23:3129–39. aneurysm progression and inhibition of galectin-3 decreases elastase-induced AAA [20] Delbosc S, Diallo D, Dejouvencel T, Lamiral Z, Louedec L, Martin-Ventura J-L, et al. development. Clin Sci (Lond) 2017;131:2707–19. Impaired high-density lipoprotein anti-oxidant capacity in human abdominal aortic [36] Escolà-Gil JC, Lee-Rueckert M, Santos D, Cedó L, Blanco-Vaca F, Julve J. Quantification aneurysm. Cardiovasc Res 2013;100:307–15. of in vitro macrophage cholesterol efflux and in vivo macrophage-specific reverse [21] Burillo E, Tarin C, Torres-Fonseca M-M, Fernandez-García C-E, Martinez-Pinna R, cholesterol transport. Methods Mol Biol 2015;1339:211–33. Martinez-Lopez D, et al. Paraoxonase-1 overexpression prevents experimental ab- [37] DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more dominal aortic aneurysm progression. Clin Sci 2016;130:1027–38. correlated receiver operating characteristic curves: a nonparametric approach. Bio- [22] Martínez-López D, Cedó L, Metso J, Burillo E, García-León A, Canyelles M, et al. Im- metrics 1988;44:837–45. paired HDL (high-density lipoprotein)-mediated macrophage cholesterol efflux in [38] Huang Y, DiDonato JA, Levison BS, Schmitt D, Li L, Wu Y, et al. An abundant patients with abdominal aortic aneurysm-brief report. Arterioscler Thromb Vasc dysfunctional apolipoprotein A1 in human atheroma. Nat Med 2014;20:193–203. Biol 2018;38:2750–4. [39] Martinez-Pinna R, Lindholt JS, Madrigal-Matute J, Blanco-Colio LM, Esteban-Salan M, [23] Zheng L, Nukuna B, Brennan M-L, Sun M, Goormastic M, Settle M, et al. Apolipopro- Torres-Fonseca MM, et al. From tissue iron retention to low systemic haemoglobin tein A-I is a selective target for myeloperoxidase-catalyzed oxidation and functional levels, new pathophysiological biomarkers of human abdominal aortic aneurysm. impairment in subjects with cardiovascular disease. J Clin Invest 2004;114:529–41. Thromb Haemost 2014;112:87–95. [24] Bergt C, Pennathur S, Fu X, Byun J, O'Brien K, McDonald TO, et al. The [40] Smith CK, Vivekanandan-Giri A, Tang C, Knight JS, Mathew A, Padilla RL, et al. Neu- myeloperoxidase product hypochlorous acid oxidizes HDL in the human artery trophil extracellular trap-derived enzymes oxidize high-density lipoprotein: an ad- wall and impairs ABCA1-dependent cholesterol transport. Proc Natl Acad Sci U S A ditional proatherogenic mechanism in systemic lupus erythematosus. Arthritis 2004;101:13032–7. Rheumatol 2014;66:2532–44. [25] Bagwan N, Bonzon-Kulichenko E, Calvo E, Lechuga-Vieco AV, Michalakopoulos S, [41] Peng D-Q, Brubaker G, Wu Z, Zheng L, Willard B, Kinter M, et al. Apolipoprotein A-I Trevisan-Herraz M, et al. Comprehensive quantification of the modified proteome tryptophan substitution leads to resistance to myeloperoxidase-mediated loss of reveals oxidative heart damage in mitochondrial Heteroplasmy. Cell Rep 2018;23: function. Arterioscler Thromb Vasc Biol 2008;28:2063–70. 3685–97. [42] Rohatgi A, Khera A, Berry JD, Givens EG, Ayers CR, Wedin KE, et al. HDL cholesterol [26] Fontaine V, Touat Z, Mtairag EM, Vranckx R, Louedec L, Houard X, et al. Role of leu- efflux capacity and incident cardiovascular events. N Engl J Med 2014;371: kocyte elastase in preventing cellular re-colonization of the mural thrombus. Am J 2383–93. Pathol 2004;164:2077–87. [43] Shao B, Tang C, Sinha A, Mayer PS, Davenport GD, Brot N, et al. Humans with athero- [27] Burillo E, Jorge I, Martínez-López D, Camafeita E, Blanco-Colio LM, Trevisan-Herraz sclerosis have impaired ABCA1 cholesterol efflux and enhanced high-density lipo- M, et al. Quantitative HDL proteomics identifies Peroxiredoxin-6 as a biomarker of protein oxidation by myeloperoxidase. Circ Res 2014;114:1733–42. human abdominal aortic aneurysm. Sci Rep 2016 09;6:38477.