RESEARCH ARTICLE

A comprehensive study of calcific aortic stenosis: from rabbit tohuman samplesLaura Mourino-Alvarez1, Montserrat Baldan-Martin1, Tamara Sastre-Oliva1, Marta Martin-Lorenzo2,Aroa Sanz Maroto2, Nerea Corbacho-Alonso1, Raul Rincon1, Tatiana Martin-Rojas1,Luis Fernando Lopez-Almodovar3, Gloria Alvarez-Llamas1, Fernando Vivanco1, Luis Rodriguez Padial4,Fernando de la Cuesta5 and Maria Gonzalez Barderas1,*

ABSTRACTThe global incidence of calcific aortic stenosis (CAS) is increasingowing, in part, to a growing elderly population. The condition posesa great challenge to public health, because of the multiplecomorbidities of these older patients. Using a rabbit model of CAS,we sought to characterize protein alterations associated with calcifiedvalve tissue that can be ultimately measured in plasma as non-invasive biomarkers of CAS. Aortic valves from healthy and mildstenotic rabbits were analyzed by two-dimensional difference gelelectrophoresis, and selected reaction monitoring was used todirectly measure the differentially expressed proteins in plasmafrom the same rabbits to corroborate their potential as diagnosticindicators. Similar analyses were performed in plasma from humansubjects, to examine the suitability of these diagnostic indicatorsfor transfer to the clinical setting. Eight proteins were found tobe differentially expressed in CAS tissue, but only three were alsoaltered in plasma samples from rabbits and humans: transitionalendoplasmic reticulum ATPase, tropomyosin α-1 chain and L-lactatedehydrogenase B chain. Results of receiver operating characteristiccurves showed the discriminative power of the scores, whichincreased when the three proteins were analyzed as a panel. Ourstudy shows that a molecular panel comprising three proteins relatedto osteoblastic differentiation could have utility as a serum CASindicator and/or therapeutic target.

KEY WORDS: Aortic stenosis, Cardiovascular, Proteomics,Rabbit model

INTRODUCTIONAortic stenosis is defined as a narrowing of the aortic valve (AV),which results in reduced blood flow to the body and ultimately incompromised heart function (Rajamannan et al., 2011). Calcificaortic stenosis (CAS), the most common etiology of aortic stenosis

in Western countries, is characterized by an inflammatory processand endothelial damage caused by mechanical stress and lipidpenetration, leading to fibrosis and leaflet thickening (Dweck et al.,2012). As the disease progresses, matrix remodeling and active boneformation occurs, ultimately leading to calcification (Otto, 2006).CAS has a prolonged asymptomatic period defined as aorticsclerosis, during which time calcification of the valve beginsto occur, but with no elevation of the transvalvular gradient.Nevertheless, once symptoms develop, CAS is rapidly fatal as thereis no effective pharmacologic treatment (Dweck et al., 2012).Patient management includes balloon valvuloplasty, which only hastransient effects, and aortic valve replacement, either surgical orusing transcatheter aortic valve implantation (TAVI) (Joseph et al.,2016). Accordingly, there is a great unmet need for alternativetherapies to reduce the overall burden of this disease. Efforts havebeen directed at controlling CAS progression and towards betterunderstanding the molecular mechanisms of CAS to providepotential indicators at early stages of the disease.

CAS is a multifactorial disease and an important challenge in itsstudy is the presence of comorbidities, including its increasedincidence with age. Animal models have been instrumental indissecting the pathogenesis of CAS, as they allow a perfect controlof external factors. In this respect, the rabbit model of aortic stenosishas been particularly useful, because of the similarities betweenrabbits and humans in terms of valve histology and lipoproteinmetabolism (Cimini et al., 2005; Turk and Laughlin, 2004).Moreover, the existence of osteogenic cells in pathological valveshas been previously described in both humans and rabbits (Kapustinet al., 2011; New and Aikawa, 2013; Liberman et al., 2008; Droletet al., 2008).

We have previously applied different strategies to investigate themolecular changes taking place during CAS using diverse biologicalsamples such as plasma (Mourino-Alvarez et al., 2016a; Gil-Doneset al., 2012), the secretome (Alvarez-Llamas et al., 2013) and tissue(Mourino-Alvarez et al., 2016b; Martín-Rojas et al., 2016, 2012). Inthis work, we have directly analyzed AV tissue from mild stenoticrabbits using a proteomic approach to identify alterations at themolecular level, which might also be reflected in plasma.

RESULTSDevelopment of CAS in the rabbit modelThe occurrence of CAS was evaluated by transthoracicechocardiographic examination (Table S1). As expected, after12 weeks on a cholesterol-enriched chow plus vitamin D2 diet, allrabbits from the pathological group (n=7) showed higher peakgradient and thickened AVs than the control group (n=7; Fig. 1A,B),which confirmed the development of CAS. By contrast, theinterventricular septum and left ventricular free wall, as well asReceived 22 December 2017; Accepted 3 May 2018

1Department of Vascular Physiopathology, Hospital Nacional de Paraplejicos,SESCAM, 45071 Toledo, Spain. 2Department of Immunology, IIS-FundacionJimenez Diaz, 28040Madrid, Spain. 3Cardiac Surgery, Hospital Virgen de la Salud,SESCAM, 45071 Toledo, Spain. 4Department of Cardiology, Hospital Virgen de laSalud, SESCAM, 45071 Toledo, Spain. 5Centre for Cardiovascular Science,University of Edinburgh, Queen’s Medical Research Institute, EdinburghEH16 4TJ, UK.

*Author for correspondence (megonzalezb@sescam.jccm.es)

L.M., 0000-0002-3928-1554; M.B., 0000-0003-0848-8154; F.d.l.C., 0000-0001-8761-7713; M.G.B., 0000-0003-4290-4721

This is an Open Access article distributed under the terms of the Creative Commons AttributionLicense (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use,distribution and reproduction in any medium provided that the original work is properly attributed.

1

© 2018. Published by The Company of Biologists Ltd | Disease Models & Mechanisms (2018) 11, dmm033423. doi:10.1242/dmm.033423

Disea

seModels&Mechan

isms

left ventricular function (reflected by ejection fraction), were notsignificantly modified by the diet (Table S1).Blood analyses were performed at the beginning and during the

development of the study, after 6 and 12 weeks of controlled diet(Table S2 and Fig. 1C). Results showed that levels of totalcholesterol were significantly higher in the pathological group thanin the control group (13.99±0.29 vs 0.53±0.17 g/l; P<0.001).Specifically, the pathological/control ratio of low-density lipoprotein(LDL) cholesterol was >58 (13.5±0.55 vs 0.23±0.14 g/l; P<0.001).Significant differences were also found between pathological andcontrol groups for high-density lipoprotein (HDL) and non-HDLcholesterol levels at the time of sacrifice (P<0.05).Histological analysis of AVs from the pathological group

revealed the presence of moderate calcium deposits (as shown byAlizarin Red staining), abundant infiltration of macrophages(RAM11-positive cells 2.09±1.61% in the pathological group vs0.015±0.014% in controls; P=0.023) and high expression ofα-actin (0.91±0.74% in the pathological group vs 0.015±0.014%in controls; P=0.018), which is characteristic of smooth musclecells and myofibroblasts (Fig. 2). In addition to differences atthe histological level, we noted that the pathological grouphad thicker AVs than the control group, as seen in theechocardiographic examination.

Two-dimensional difference gel electrophoresis analysisand differentially expressed proteinsWe used two-dimensional difference gel electrophoresis (2D-DIGE)in combination with tandem mass spectrometry to compare therelative abundance of proteins extracted from valve tissue in the twogroups. Scanned gel images were analyzed withDeCyder DifferentialAnalysis Software (GEHealthcare, Chicago, IL, USA), which allowsdetection, quantitation, matching and statistical analysis of the

images. Statistical analysis (Student’s t-test) revealed significantchanges in the abundance (P≤0.05 and average ratio >1.5 or <−1.5)of 15 spots: five were upregulated in CAS tissue and ten weredownregulated (Fig. 3). The results were analyzed using PrincipalComponent Analysis (PCA) to reduce the complexity of the data setand to look for distinctive proteome profiles in the two study groups.As shown in Fig. S1, pathological and control valves were separatedinto two different groups.

Matrix-assisted laser desorption/ionization time-of-flight tandemmass spectrometry (MALDI TOF/TOF) was used to identifythe significantly altered spots, which revealed eight proteins.Calreticulin, transglutaminase-2 and transitional endoplasmicreticulum ATPase (TERA) were upregulated in CAS, whereas serumalbumin, tropomyosin α-1 chain (TPM-1), L-lactate dehydrogenase Bchain (LDHB),myosin light chain 3 andmyosin regulatory light chain2, ventricular/cardiac muscle isoform were downregulated (Table 1).

As myosins, TPM-1 and LDHB are highly expressed inmyocardial tissue (Kamakura et al., 2013; Lossie et al., 2014; Liuet al., 2016), we confirmed their presence in valve tissue byimmunohistochemistry (Fig. S2).

Selected reaction monitoringWe were able to monitor five of the proteins identified in rabbitplasma by selected reaction monitoring (SRM) using liquidchromatography tandem-mass spectrometry (LC-MS/MS).Among them, three had a consistent result in the three transitionsof the two peptides that were measured: TERA, TPM-1 and LDHB.TPM-1 and LDHB were found to be significantly downregulated inplasma, whereas TERA was significantly upregulated (Table 2 andFig. 4).We also measured the levels of these proteins in plasma frompatients with AV disease (n=34) and from control subjects (n=12)(characteristics of the patient study groups are shown in Table 3).

Fig. 1. The evaluation of CAS in the rabbit model. (A,B) Representative echocardiograms from controls (A) and pathological rabbits (B) after 12 weeks of diet.Doppler velocity is shown in the upper images (blue arrows), whereas the aortic valves are shown by white arrows in the lower images. It can be observed thatpathological rabbits have higher Doppler velocity (and, subsequently, higher transvalvular gradient) and thicker aortic valves. Aorta, left ventricle (LV) and right atrium(RA) are indicated in the figure. (C) Results from blood analyses, with significant differences marked: *P<0.05, ***P<0.001. P-values were calculated by comparingeach corresponding time point to t=0 using a paired Student’s t-test. Red lines, pathological group; green lines, control group.

2

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033423. doi:10.1242/dmm.033423

Disea

seModels&Mechan

isms

This analysis showed that TERA, TPM-1 and LDHB were alsosignificantly altered in human subjects and followed the same trendas the rabbit plasma (Table 2 and Fig. 4).Results from rabbit and human plasma were used to assess the

sensitivity and specificity of these potential markers by individualreceiver operating characteristic (ROC) curves. In rabbits, the areaunder the curve (AUC) was 1.0 with significant values (P<0.01) forall peptides. In human plasma, the AUC was higher than 0.73 andthe P-value below 0.037 in all cases. Moreover, when the threeproteins were combined, the AUC increased to 1.0 and the P-valuedecreased to 6.28×10−6 (Fig. 5).

DISCUSSIONAnimal models are of great utility for studying several diseases, asthey allow the control of differences between experimental groups.This is especially important in the setting of CAS given the etiologyof the disease: at advanced ages, CAS is usually accompanied byother comorbidities such as hypertension or diabetes. Our rabbitmodel was based on a hypercholesterolemic diet supplemented withvitamin D2, which has been shown to be effective both for the studyof the evolution of the early phases of AV disease (Drolet et al.,2003) and for investigating the potential prevention of progressionto severe stages (Busseuil et al., 2008; Drolet et al., 2003). Theincrease in the transvalvular gradient confirmed the development ofCAS in the pathological group, which was concomitant with asignificant increase in cholesterol (total, LDL, HDL and non-HDL),characteristics that have been previously described in patients withCAS (Kamath and Rai, 2008; Akat et al., 2010). By contrast, theechocardiography study showed no development of hypertrophy orimpaired systolic function of the left ventricle, which is consistent

with development of mild CAS rather than severe disease(Seiler and Jenni, 1996; Mihaljevic et al., 2008). Importantly, theanimal model allowed us to study valve tissue against a backgroundof mild calcification levels. It is difficult to obtain mild calcifiedvalves from patients, as AV replacement is recommended only atadvances stages of CAS when symptoms appear owing to severedamage in valve mobility.

Our protein analysis revealed eight proteins of interest of whichthree were significantly altered, as shown by SRM analysis ofplasma both from rabbits and patients: TPM-1 and LDHB followedthe same trend in plasma and tissue, whereas TERAwas upregulatedin tissue and downregulated in both rabbit and human plasma.

Structural proteins such as TPM-1 are part of the cardiac muscleand the contractile cytoskeleton of various cell types, includingfibroblasts and endothelial cells, and are essential for themaintenance of the endothelial barrier (Mehta and Malik, 2006).Overexpression of TPM-1 has been shown to stabilize the structureof actin filaments, helping to preserve endothelial barrier functionunder conditions of oxidative stress (Gagat et al., 2014). Moreover, areduction in the expression of TPM-1 has been previously related toa decrease in the contraction of actin filaments in arteries witharteriosclerosis (Wang et al., 2011). The reduction of TPM-1 levelsmight also relate to the loss of AV flexibility, as occurs inarteriosclerosis, being indicative of the differentiation of thevalvular interstitial cells to osteoblasts, which have no contractileability (Lei et al., 2014; Park et al., 2009).

Valve calcification is also related to higher endoplasmic reticulumstress (ERS) (Cai et al., 2013). In a previous study, we showed thatoxidized LDL promotes osteoblastic differentiation through theactivation of the ERS pathway. In the present study, we found higher

Fig. 2. Histology of the aortic valves from pathological and control rabbits. (A,B) Hematoxylin and Eosin (H&E) staining reveals increased valve thickness inpathological rabbits. (C,D) Alizarin Red staining highlights the presence of calcium deposits in pathological rabbits (arrow). (E-J) Macrophage (RAM11; E-G)and SMC and/or myofibroblast (α-actin; H-J) staining is more intense in the pathological group (arrows). Scale bars: 200 µm (100× images); scale bars: 50 µm(400× images). Asterisks indicate the aortic valve.

3

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033423. doi:10.1242/dmm.033423

Disea

seModels&Mechan

isms

levels of TERA, also named valosin-containing protein, in the AVtissue of the pathological group. TERA is responsible for exportingmisfolded proteins to the endoplasmic reticulum, where theyaccumulate and trigger ERS (Jarosch et al., 2002). In a similar wayto TPM-1, the increase of TERA found in our study points to thedifferentiation of valvular interstitial cells to osteoblasts, which hasbeen previously related to ERS (Cai et al., 2013).In valve tissue, oxygen requirements exceed the amount

deliverable by diffusion from the cusp surfaces alone, thus

vasculature is needed to maintain a sufficient oxygen supply tothe cells (Weind et al., 2000, 2002). It is therefore reasonable toassume that tissue thickening might lead to a reduction in theamount of oxygen received by these cells. In hypoxic conditions,lactate is formed from pyruvate by L-lactate dehydrogenase A chain(LDHA). LDHB is a heart-specific isoform that catalyzes the samereaction in aerobic conditions (Buono and Lang, 1999). Thisisoform has been shown to be downregulated in conditions ofoxygen deprivation (Rossignol et al., 2003; Kay et al., 2007), which

Table 1. Proteins identified by mass spectrometry (MALDI TOF/TOF) that were found to show significantly different levels in 2D-DIGE analysis

Accessionnumber Name

CAS/control P-value Function MASCOT

Sequencecoverage (%)

Matchedpeptides (n)

1 P15253 Calreticulin 2.38 0.013 Chaperone 260 82.2 262 G1SK42 Protein-glutamine

γ-glutamyltransferase 21.85 0.025 Acyltransferase 464 45.4 33

3 G1SR03 Transitional endoplasmic reticulumATPase

1.67 0.038 Hydrolase 175 42.4 26

4 P49065 Serum albumin −1.74 0.011 Antioxidant/Transport

318 67.1 33

8 P58772 Tropomyosin α-1 chain −5.28 0.046 Muscular protein 85 34.5 910 P13490 L-Lactate dehydrogenase B chain −1.67 0.023 Oxidoreductase 94 51.2 1314 G1T375 Myosin light chain 3 −5.41 0.040 Muscular protein 196 68.4 1415 Q7M2V4 Myosin regulatory light chain 2,

ventricular/cardiac muscle isoform−6.03 0.041 Muscular protein 93 53.9 11

The table shows spot number, accession number according to UniProt, protein name, statistical results (ratio CAS/control and P-value according to Student’st-test), main function, score obtained in the identification using MASCOT, sequence coverage and number of matched peptides.

Fig. 3. Master gel of rabbit valve 2D-DIGE showing 15 differential protein spots between control and pathological groups. Spots that are increased in thepathological group are shown in red, whereas spots that are decreased in this group are shown in green.

4

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033423. doi:10.1242/dmm.033423

Disea

seModels&Mechan

isms

is in accordance with our results in CAS tissue and in serum fromrabbits and humans with CAS.We believe that the alterations we have found at the tissue level

are significant, as they provide molecular information about themechanisms that take place within the valve. Moreover, theseproteins might serve as potential therapeutic targets to slow downthe progression of the disease, although more functional analysis isneeded. Of particular interest is the opposite trend we found forexpression of TERA in tissue (higher in stenotic valves) and plasma(lower levels in CAS rabbits/subjects). TERA is commonly found inextracellular vesicles in several types of cells, including endothelialcells (Peterson et al., 2008) and lymphocytes (Miguet et al., 2006).TERA has also been identified in microparticles derived fromhuman atherosclerotic plaques (Mayr et al., 2009). Given theincreasing importance of extracellular vesicles in cardiovasculardisease (Yin et al., 2015; Chen et al., 2018; Badimon et al., 2017),these variations in TERA levels should be further studied as theycould be indicative of differential tissue and/or cell vesicle releaseduring the development of CAS.We have demonstrated the great potential of SRM to quantify

differences in proteins across multiple samples. In rabbit samples,peptides from TPM-1 and LDHB do not have the same ratioestimations, probably because two peptides are outside the linearrange of the assay (Bao et al., 2013). Special care was taken to avoidthis effect in plasma samples, something crucial for the evaluationof their potential utility as diagnostic markers of these proteins. Asshown in the ROC curves, each of the verified proteins has sufficientsensitivity and specificity to discriminate between control and

pathological subjects (AUC>0.73), pointing to their potential utilityas diagnostic markers. It should be noted that the combinedmeasurement of the three proteins as a panel has greaterdiscriminatory power (AUC=1.0), presenting a very high capacityto assign subjects to their corresponding study groups. Analysis inplasma samples might facilitate the translation of this panel to theclinical field, as blood samples are easy to obtain and are minimallyinvasive compared with biopsies and surgical procedures. Our useof LC-MS/MS for verification has not been by chance. The routineuse of this technology in clinical laboratories has witnessedunprecedented growth during the past two decades owing notonly to its high specificity, sensitivity and high-throughputpotential, but also because it is faster and more flexible thanclassical immunoassays (Grebe and Singh, 2011; Leung and Fong,2014). Therefore, it is foreseeable that LC-MS/MS will become apowerful tool in routine clinical laboratory testing.

Some limitations of this work should be highlighted. One of themost important challenges of using animal models is the notablespecies differences between animal models and humans.Nevertheless, CAS in vitamin D2 supplementation models ishistologically and hemodynamically similar to the human disease,involving fibrosis/calcification, inflammatory response andendothelial dysfunction (Ngo et al., 2008), as well as changes incardiac function, as shown here. Also, the rabbit model has a greatertranslational strength than murine models. Clearly, the analysis ofhuman samples has the disadvantage that the underlyingpathological and physiological conditions cannot be controlled,especially when studying elderly patients as they present more

Table 2. Results from plasma analyses using SRM

Species Protein Peptide sequence Fragment ion Control mean±s.e.m. CAS mean±s.e.m. P-value

Rabbit TPM-1 LVIIESDLER y5 1.27±0.17 0.57±0.13 0.003y6 1.27±0.17 0.60±0.10 0.005y9 1.28±0.21 0.64±0.11 0.012

SIDDLEDELYAQK y5 1.01±0.04 0.11±0.03 <0.001y6 1.05±0.07 0.13±0.03 <0.001y8 1.05±0.08 0.13±0.03 <0.001

TERA GDDLSTAILK y3 1.38±0.15 0.74±0.08 0.003y4 1.27±0.15 0.61±0.06 0.002y6 1.17±0.10 0.69±0.08 0.003

MDELQLFR y2 1.33±0.16 0.64±0.07 0.002y3 1.27±0.14 0.68±0.11 0.005y5 1.39±0.17 0.69±0.11 0.004

LDHB FIIPQIVK y3 1.06±0.18 0.58±0.10 0.023y5 1.00±0.11 0.59±0.12 0.020y6 1.08±0.15 0.68±0.13 0.039

MVVESAYEVIK y7 4.45±1.43 0.84±0.13 0.032y8 6.16±3.65 0.65±0.08 0.102y9 4.59±1.61 0.94±0.21 0.043

Human TPM-1 QLEDELVSLQK y8 8.58E-04±1.26E-04 4.73E-04±4.22E-05 0.006y6 4.09E-04±8.90E-05 3.00E-04±3.12E-05 0.149y5 5.05E-04±7.84E-05 3.06E-04±3.24E-05 0.007

TERA LEILQIHTK y6 4.37E-03±6.52E-04 2.66E-03±1.95E-04 0.013b5 1.04E-02±1.52E-03 6.37E-03±4.51E-04 0.013b6 5.07E-03±7.32E-04 3.12E-03±2.24E04 0.012

GGNIGDGGGAADR y7 1.74E-02±3.66E-03 9.12E-03±8.08E-04 0.024b9 1.06E-03±2.16E-04 5.58E-04±5.12E-05 0.021b12 4.60E-03±9.39E-04 2.46E-03±2.27E-04 0.023

LDHB MVVESAYEVIK y9 3.82E-02±6.33E-03 2.24E-02±1.60E-03 0.016y8 1.18E-02±1.96E-03 6.91E-03±4.76E-03 0.016b8 5.08E-02±7.97E-03 3.00E-02±2.04E-03 0.013

GLTSVINQK b6 2.64E-02±4.20E-03 1.69E-02±1.22E-03 0.024b7 1.30E-01±2.05E-02 8.32E-02±6.06E-03 0.024b8 9.18E-03±1.41E-03 5.95E-03±4.30E-04 0.023

Shown are the peptides and transitions measured for each protein and the statistical analyses for each transition, including mean and P-value.

5

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033423. doi:10.1242/dmm.033423

Disea

seModels&Mechan

isms

concomitant diseases. Finally, according to the classicaldevelopment biomarker pipeline (Surinova et al., 2011), theseproteins should be validated in an independent cohort of at least 100subjects prior to clinical evaluation.In summary, we have defined a new molecular panel that can be

measured in plasma using an extremely reproducible and reliablemethod, such as SRM, indicating its potential for implementation inthe clinic. Nevertheless, it will be necessary to perform furtherstudies to assess some remaining aspects. It will be important to usein vitro and in vivomodels to clearly define the role of these proteinsin tissue. If, as we hypothesize, three of these proteins are relatedto osteoblastic differentiation, they might represent putativetherapeutic targets to reduce calcium deposition. In addition,studies on the implications of ERS in valve calcification arescarce and it would be interesting to further our understanding ofthese mechanisms and to study the specific role of TERA in AVcalcification. Finally, a prospective study in a larger cohort ofsubjects with different degrees of AV damage, frommild sclerosis to

severe stenosis, should be carried out to validate the diagnostic andprognostic value of these indicators. If these candidates are suitablefor clinical evaluation, it might lead to a considerable improvementin patient management, reducing the burden of CAS in society.

MATERIALS AND METHODSAnimal modelMale New Zealand white rabbits (Oryctolagus cuniculus) weighing2–2.5 kg were divided into two groups: animals in the control group(n=7) were fed with normal rabbit chow; animals in the pathological group(n=7) were fed with 1% cholesterol-enriched chow plus 50,000 IU/kgvitamin D2 (Harlan, Indianapolis, IN, USA). All animals were fedad libitum for 12 weeks (Drolet et al., 2008). Echocardiographicevaluations of the AV were performed at t=0, t=6 weeks and t=12 weeks,to ensure the establishment of CAS. Blood was drawn into EDTA tubesthrough the marginal vein of the ear simultaneously for the measurement ofcholesterol and triglycerides. After the 12-week period, animals weresedated with an injection of ketamine (100 mg/kg) and xylazine (20 mg/kg)and then euthanized by injection of pentobarbital (50 mg/kg) directly intothe heart. AVs were immediately harvested, rinsed in saline buffer andprocessed for analyses. When the analysis was not performed immediately,tissues were stored at −80°C.

The study was conducted in accordance with the Principles of LaboratoryAnimal Care and all experimental procedures were approved by the AnimalCare and Use Committee of the IIS-Fundación Jiménez Díaz, according tothe guidelines for ethical care of the European Community.

EchocardiographyUltrasound video images were obtained using the HD11XEechocardiographic system (Philips Medical Imaging, Best, TheNetherlands) and a neonatal S12-4 ultrasound imaging probe, with anextended frequency range of 4–12 MHz. A parasternal long-axis view was

Fig. 4. Plasma analysis using SRM. (A,B) Plasma analysis was performed in rabbit (A) and human (B) samples. SRM analyses allowed the measurement oftropomyosin α-1 (TMP-1), transitional endoplasmic reticulum ATPase (TERA) and L-lactate dehydrogenase B chain (LDHB). All the transitions were used tocalculate the mean intensity of each peptide. Relative abundance is shown (100% corresponds to control group).

Table 3. Clinical characteristics of the subjects used in the validationphase with human samples

Control (n=12) CAS patients (n=34) P-value

Mean age (years±s.d.) 65.0±21.96 74.9±8.10 0.47Gender, M/F (%) 58/42 50/50 0.74Obesity (%) 8 26 0.25AHT (%) 58 82 0.12Dyslipidemia (%) 25 59 0.09Diabetes (%) 17 32 0.46Smoker (%) 0 21 0.17

M/F, male/female; AHT, arterial hypertension.

6

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033423. doi:10.1242/dmm.033423

Disea

seModels&Mechan

isms

used to measure valvular thickness and left ventricle parameters. Leftventricular ejection fraction and fractional shortening were calculated frommeasurements of the left ventricular internal diameter in systole anddiastole. Additionally, aortic outflow velocity was registered usingcontinuous-wave Doppler echocardiograpy from apical planes, and thepeak gradient was calculated using the Bernoulli equation (Baumgartneret al., 2009).

Tissue stainingOne leaflet of each valve was fixed in 4% buffered-formalin for 24 h andthen embedded in paraffin. Paraffin-embedded sections were subjectedto Hematoxylin and Eosin (H&E) and Alizarin Red staining forvisualization of calcium deposits. Six monoclonal antibodies wereused for immunohistochemistry: RAM11 for macrophages (dilution1:100; M0633; DAKO, Santa Clara, CA, USA), α-actin for vascular smoothmuscle cells (1:100; M0851; DAKO), tropomyosin α-1 (1:25; sc-376541;Santa Cruz Biotechnology, Dallas, TX, USA), L-lactate dehydrogenase Bchain (1:200; sc-100775; Santa Cruz Biotechnology), myosin light chain 3(1:2000; ab680; Abcam, Cambridge, UK) and myosin regulatory light chain2 (1:200; ab89594; Abcam). In control experiments, no primary antibodywas added. Non-specific binding was prevented by incubation with normalgoat serum (for RAM11) or 10% bovine serum albumin (for the remainder)for 1 h; non-specific peroxidase activity was blocked by incubation with 3%hydrogen peroxidase for 5 min. Incubation with primary antibodies wasperformed for 1 h at room temperature. The slides were then incubated withhorseradish-peroxidase-conjugated polyclonal goat anti-mouse antibodies(dilution 1:100; P0447; DAKO) for 30 min, and the chromogenic reactionwas developed using 3,3′-diaminobenzidine (DAB). Sections werecounterstained with Hematoxylin prior to dehydration and the addition of

a coverslip. For an impartial analysis of the DAB staining, an orthonormaltransformation of the RGB images using an ImageJ plugin (NIH) based onRuifrok and Johnston’s method for color deconvolution was performed(Ruifrok and Johnston, 2001).

Proteomic analysis using two-dimensional difference gelelectrophoresisOne AV leaflet was ground into powder in liquid nitrogen with a mortar.Proteins were then extracted using 7 M urea, 2 M thiourea, 4% CHAPS and1% dithiothreitol (DTT) and the homogenate was centrifuged to precipitatetissue debris. The supernatant was collected and the protein concentrationdetermined using the Bradford assay.

Before proteomic analysis, the required amount of protein was subjectedto a cleaning step by precipitation using the 2D Clean-Up kit (GEHealthcare, Chicago, IL, USA) and resuspended in rehydratation buffer(7 M urea, 2 M thiourea, 4% CHAPS, 30 mM Tris) to a final concentrationof 7 mg/ml. Proteins were then labeled according to the manufacturer’sinstructions (GE Healthcare). Briefly, 50 µg of protein from each AV extractwas labeled with 400 pmol of N-hydroxysuccinimide esters of Cy3 or Cy5fluorescent cyanine dye for 30 min on ice in the dark. An internal standardcontaining equal amounts of all experimental samples was labeledwith 400 pmol of N-hydroxysuccinimide Cy2 dye. Reactions were thenquenched with 0.2 mM lysine. Labeled protein extracts were combinedaccording to the experimental design (Table S3), diluted in rehydrationbuffer (7 M urea, 2 M thiourea, 4%CHAPS, 30 mMTris) with 2%DTT and1% ampholytes (IPG buffer pH 4–7, GE Healthcare) and applied to 24 cmpH 4–7 IPG strips. After passive rehydration, the first dimension was run onthe IPGphor IEF II System (GE Healthcare), as described: 500 V for 1 h, alinear gradient to 1000 V over 2 h, a linear gradient to 8000 V over 3 h, and

Fig. 5. Assessment of the sensitivity and specificity of potential markers using ROC curves. ROC curves of tropomyosin α-1 (TMP-1), transitionalendoplasmic reticulum ATPase (TERA) and L-lactate dehydrogenase B chain (LDHB) are shown in the upper part of the figure. ROC curve of the combinedproteins is shown in the lower image. When these proteins are combined, the obtained panel is more sensitive and specific, thus it would be more useful thanindividual proteins for the development of clinical diagnostic tools. In all cases, the transition of the most significant peptide is represented. Area under the curve(AUC) and P-values are shown.

7

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033423. doi:10.1242/dmm.033423

Disea

seModels&Mechan

isms

8000 V until 96,000 V/h. After the first dimension, the strips wereequilibrated in SDS equilibration buffer (1.5 M Tris-HCl pH 8.8, 6 Murea, 87% glycerol and 2% SDS) using a two-step protocol for reduction(by adding 1% DTT) and alkylation (by adding 2.5% iodoacetamide) of thiolgroups. Proteins were then separated on 10% acrylamide/bisacrylamide gelsusing an EttanDalt Six device (GE Healthcare) (Laemmli, 1970).

Image acquisition and analysisGels were scanned on a Typhoon 9400 fluorescence gel scanner(GE Healthcare) using appropriate individual excitation and emissionwavelengths, filters and photomultiplier values sensitive for each of the Cy3,Cy5 and Cy2 dyes.

Images were analyzed using DeCyder software v6.5 (GE Healthcare).The Differential In-gel Analysis module co-detected the three images ofeach gel (spot maps from the internal standard and the two samples),measured the spot abundance in each image, and expressed these values asCy3/Cy2 and Cy5/Cy2 ratios. The Biological Variation Analysis moduleenabled the matching of these spot maps, the comparison of the Cy3/Cy2and Cy5/Cy2 ratios and the statistical analysis, to determine changes inexpression levels. Only protein spots with >1.5-fold difference in abundanceand with P-values below 0.05 (Student’s t-test) were considered as proteinsof interest. Finally, a multivariate analysis was performed by PCA usingthe Extended Data Analysis module. A pattern analysis hierarchicalclassification was also obtained using the Pearson coefficient based onthe spots present in 90% of all the gels.

In-gel digestion and protein identification by matrix-assistedlaser desorption/ionization time-of-flight mass spectrometryDifferentially expressed protein spots were excised manually from the2D-DIGE gels, which were previously stained with OrioleTM fluorescent gelstain (Bio-Rad, Hercules, CA, USA). Additionally, a preparative gel using400 µg of total protein was prepared for the identification of small or low-abundance proteins, using the same electrophoretic parameters. Spots wereautomatically digested with the Ettan Digester workstation (GE Healthcare)and identified at the Proteomic Unit of Hospital Nacional de Parapléjicos.The digestion was performed according to Shevchenko et al. (1996) withminor modifications and, after digestion at 37°C overnight, the peptideswere extracted with 60% acetonitrile (ACN) in 0.1% formic acid. Sampleswere dried in a speedvac and resuspended in 98% water with 0.1% formicacid and 2% ACN. An aliquot of each digestion was mixed with an aliquotof the matrix solution (3 mg/ml matrix α-cyano-4-hydroxycinnamic acid in30% ACN, 15% 2-propanol and 0.1% trifluoroacetic acid), and this waspipetted directly on a 384 Opti-TOF 123×81 mm stainless steel sample plateand analyzed on a 4800 Plus MALDI TOF/TOF mass spectrometer(Applied Biosystems, Foster City, CA, USA).

MALDI-MS/MS analysis and database searchingMALDI-MS/MS data were obtained using an automated analysis loop inthe MALDI TOF/TOF Analyzer. MALDI-MS and MS/MS data werecombined using GPS Explorer Software Version 3.6 to search a non-redundant protein database (Swissprot 56.5) with Mascot software version2.2 (Matrix Science, London, UK) (Perkins et al., 1999) applying theappropriate settings: 50 ppm precursor tolerance, 0.6 DaMS/MS fragmenttolerance, one missed cleavage allowed, carbamidomethyl cysteines andmethionine oxidation as modifications. The MALDI-MS/MS spectra anddatabase search results were manually inspected in detail using theaforementioned software.

Patient selection and blood extractionPeripheral blood samples were collected from control subjects (n=12) andpatients with severe CAS (n=34) who underwent scheduled AV replacementat Hospital Virgen de la Salud (Toledo, Spain). In patients, AV area(0.74±0.19 cm2), ejection fraction (57.13±9.83%) and mean gradient(48.16±18.06 mmHg) were assessed using transthoracic echocardiography.Blood samples were always taken before surgery. Subjects were selectedto avoid significant differences between the groups in terms of sixmain cardiovascular risk factors: age, gender, obesity, hypertension,

dyslipidemia and diabetes (Table 3). Samples from patients with bicuspidAV, concomitant aortic stenosis and aortic regurgitation or mitral valvedisease were excluded.

The patient study was carried out in accordance with the recommendationsof the Helsinki Declaration and was approved by the ethics committee atthe Hospital Virgen de la Salud. Signed informed consent was obtained fromall subjects. Blood samples (28 ml) were drawn into EDTA-containing tubesand centrifuged at 1125 g for 15 min and the resulting supernatant wasimmediately frozen at −80°C until analysis.

Selected reaction monitoringProteins from plasma samples were reduced and alkylated by incubatingwith 100 mM DTT and 550 mM iodoacetamide in 50 mM ammoniumbicarbonate, respectively. Proteins were digested in 50 mM ammoniumbicarbonate, 15% acetonitrile with sequencing grade modified porcinetrypsin, at a final concentration of 1:50 (trypsin:protein). After overnightdigestion at 37°C, 2% formic acid was added and samples were cleaned withPep-Clean spin columns (Pierce, Waltham, MA, USA). Tryptic digests weredried in a speedvac and resuspended in 2% ACN, 2% formic acid prior toMS analysis.

The LC-MS/MS system comprised a TEMPO nano LC system (AppliedBiosystems) combined with a nano LC Autosampler and coupled to amodified triple quadrupole MS system (Applied Biosystems 4000 QTRAPLC-MS/MS). Three replicate injections (4 μl containing 1 μg of protein)were made for each sample using mobile phase A (2% ACN/98% water,0.1% formic acid) with a flow rate of 10 μl/min for 5 min. Peptides wereloaded onto a μ-Precolumn Cartridge (Acclaim Pep Map 100 C18, 5 μm,100 Å; 300 μm i.d.×5 mm; LC Packings, Amsterdam, The Netherlands) topreconcentrate and desalt samples. Reverse-phase liquid chromatographywas achieved on a C18 column (Onyx Monolithic C18, 150×0.1 mm i.d.;Phenomenex, Torrance, CA, USA) in a gradient of phase A and phase B(98% ACN/2% water, 0.1% FA). Peptides were eluted at a flow rate of900 nl/min in three steps: 5–45% B for 60 min, 45–95% B for 1 min andfinally 95% B for 4 min. The column was then regenerated with 5% B for15 min. Both the TEMPO nano LC and 4000 QTRAP systems werecontrolled by Analyst Software v.1.5.2. The mass spectrometer was set tooperate in positive ion mode with ion spray voltage of 2800 V and ananoflow interface heater temperature of 80°C. Source gas 1 and curtain gaswere both set to 20, and nitrogen was applied as both curtain and collisiongases. For peptide selection, theoretical SRM transitions were designedusing Skyline software v3.1.0.7382 (MacCoss Lab, Seattle, WA, USA) andUnique Peptide function from that software was used to verify thattheoretical tryptic peptide sequences were proteotypic. A sample containinga mixture of all the proteins of interest was digested and analyzed witha MIDAS acquisition method, which combines a multiple reactionmonitoring (MRM) scan with a full MS/MS product ion scan to allowexamination of all fragment ions in the same spectrum. Peptides withsix co-eluting transitions with a signal-to-noise ratio over five wereconsidered for the analysis. Among these, the three most intensetransitions for each peptide were selected for the quantification and foroptimizing collision energy and dwell times to obtain maximumtransmission efficiency and sensitivity for each one (Table S4). Skylinesoftware was also used to calculate the peptide abundance on the basis ofpeak areas after integration.

Statistical analysisStatistical analyses were performed using SPSS 15.0 for windows software(SPSS Inc., Chicago, IL, USA). Continuous variables, such as age, areexpressed as mean±s.d. After demonstrating normal distribution of thepopulation using a Kolmogorov–Smirnov test, comparison of means wasperformed using Student’s t-test. Differences in variables in rabbits beforeand after the diet were analyzed using paired Student’s t-test. Discretevariables, such as sex or the presence/absence of risk factors, are expressedas percentages. In these cases, Fisher’s exact test was used for comparison ofthe groups. ROC curves were generated using SPSS 15.0.

For SRM, only peptides with significant P-values in at least two of thethree measured transitions were considered significant. Results from rabbitplasma are shown as t=12 weeks/t=0 ratio.

8

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033423. doi:10.1242/dmm.033423

Disea

seModels&Mechan

isms

AcknowledgementsWe thank the Proteomics Unit Hospital Nacional de Paraplejicos for assistancewith protein identification and theMicroscopy Unit Hospital Nacional de Paraplejicosfor assistance with image analysis.

Competing interestsThe authors declare no competing or financial interests.

Author contributionsConceptualization: F.d.l.C., M.G.B; Methodology: L.M.-A., F.d.l.C.; Software:L.M.-A., M.B.-M.; Validation: L.M.-A., M.B.-M, T.S.-O., N.C.-A., R.R., T.M-R.;Formal analysis: L.M.-A., M.B.-M., R.R., T.M.-R., M.G.B; Investigation: L.M.-A.,M.B.-M., T.S.-O., M.M.-L., A.S.M., N.C.-A., R.R., T.M.-R., L.F.L.-A., G.A.-L., F.V.,L.R.P., M.G.B.; Resources: L.F.L., L.R.P., M.G.B.; Writing - original draft: L.M.-A.,F.d.l.C., M.G.B.; Writing - review & editing: L.M.-A., M.B.-M., T.S.-O., N.C.-A., L.F.L.,L.P., F.d.l.C., M.G.B.; Visualization: M.M.-L., A.S.M., G.A.-L., F.V.; Supervision:F.d.l.C., M.G.B.

FundingThis work was supported by grants from Instituto de Salud Carlos III [PI11-02239,PI14-01917, PRB3 (IPT17/0019-ISCIII-SGEFI/ERDF)], FONDOS FEDER[RD06/0014/1015, RD12/0042/0071] and Sociedad Espan ola de Cardiologıa.

Data availabilityData from human SRM analysis has been uploaded to PeptidesAtlas repository(identifier PASS01173) (http://www.peptideatlas.org/PASS/PASS01173).

Supplementary informationSupplementary information available online athttp://dmm.biologists.org/lookup/doi/10.1242/dmm.033423.supplemental

ReferencesAkat, K., Kaden, J. J., Schmitz, F., Ewering, S., Anton, A., Klomfass, S.,Hoffmann, R. and Ortlepp, J. R. (2010). Calcium metabolism in adults withsevere aortic valve stenosis and preserved renal function. Am. J. Cardiol.105, 862-864.

Alvarez-Llamas, G., Martın-Rojas, T., de la Cuesta, F., Calvo, E., Gil-Dones, F.,Darde, V. M., Lopez-Almodovar, L. F., Padial, L. R., Lopez, J.-A., Vivanco, F.et al. (2013). Modification of the secretion pattern of proteases, inflammatorymediators, and extracellular matrix proteins by human aortic valve is key in severeaortic stenosis. Mol. Cell. Proteomics 12, 2426-2439.

Badimon, L., Suades, R., Arderiu, G., Pen a, E., Chiva-Blanch, G. and Padro, T.(2017). Microvesicles in atherosclerosis and angiogenesis: from bench to bedsideand reverse. Front. Cardiovasc. Med. 4, 77.

Bao, Y.,Waldemarson, S., Zhang, G.,Wahlander, A., Ueberheide, B., Myung, S.,Reed, B., Molloy, K., Padovan, J. C., Eriksson, J. et al. (2013). Detection andcorrection of interference in SRM analysis. Methods 61, 299-303.

Baumgartner, H., Hung, J., Bermejo, J., Chambers, J. B., Evangelista, A.,Griffin, B. P., Iung, B., Otto, C. M., Pellikka, P. A. and Quin ones, M. (2009).Echocardiographic assessment of valve stenosis: EAE/ASE recommendationsfor clinical practice. J. Am. Soc. Echocardiogr. 22, 1-23.

Buono, R. J. and Lang, R. K. (1999). Hypoxic repression of lactate dehydrogenase-B in retina. Exp. Eye Res. 69, 685-693.

Busseuil, D., Shi, Y., Mecteau, M., Brand, G., Kernaleguen, A.-E., Thorin, E.,Latour, J.-G., Rheaume, E. and Tardif, J.-C. (2008). Regression of aortic valvestenosis by ApoA-I mimetic peptide infusions in rabbits. Br. J. Pharmacol.154, 765-773.

Cai, Z., Li, F., Gong, W., Liu, W., Duan, Q., Chen, C., Ni, L., Xia, Y., Cianflone, K.,Dong, N. et al. (2013). Endoplasmic reticulum stress participates in aortic valvecalcification in hypercholesterolemic animals. Arterioscler. Thromb. Vasc. Biol.33, 2345-2354.

Chen, Y., Li, G. and Liu, M.-L. (2018). Microvesicles as emerging biomarkers andtherapeutic targets in cardiometabolic diseases. Genomics ProteomicsBioinformatics 16, 50-62.

Cimini, M., Boughner, D., Ronald, J., Aldington, L. and Rogers, K. (2005).Development of aortic valve sclerosis in a rabbit model of atherosclerosis: animmunohistochemical and histological study. J. Heart Valve Dis. 14, 365-375.

Drolet, M.-C., Arsenault, M. and Couet, J. (2003). Experimental aortic valvestenosis in rabbits. J. Am. Coll. Cardiol. 41, 1211-1217.

Drolet, M., Couet, J. and Arsenault, M. (2008). Development of aortic valvesclerosis or stenosis in rabbits: role of cholesterol and calcium. J. Heart Valve Dis.17, 381-387.

Dweck, M. R., Boon, N. A. and Newby, D. E. (2012). Calcific aortic stenosis: adisease of the valve and the myocardium. J. Am. Coll. Cardiol. 60, 1854-1863.

Gagat, M., Grzanka, D., Izdebska, M., Sroka, W. D., Marszałł, M. P. and Grzanka,A. (2014). Tropomyosin-1 protects endothelial cell-cell junctions against cigarette

smoke extract through F-actin stabilization in EA.hy926 cell line. Acta Histochem.116, 606-618.

Gil-Dones, F., Darde, V., Alonso-Orgaz, S., Lopez-Almodovar, L. F.,Mourino-Alvarez, L., Padial, L. R., Vivanco, F. and Barderas, M. G. (2012).Inside human aortic stenosis: a proteomic analysis of plasma. J. Proteomics75, 1639-1653.

Grebe, S. K. G. andSingh, R. J. (2011). LC-MS/MS in the clinical laboratory - whereto from here? Clin. Biochem. Rev. 32, 5-31.

Jarosch, E., Taxis, C., Volkwein, C., Bordallo, J., Finley, D., Wolf, D. H. andSommer, T. (2002). Protein dislocation from the ER requires polyubiquitinationand the AAA-ATPase Cdc48. Nat. Cell Biol. 4, 134-139.

Joseph, J., Naqvi, S. Y., Giri, J. and Goldberg, S. (2016). Aortic stenosis:pathophysiology, diagnosis and therapy. Am. J. Med. 130, 253-263.

Kamakura, T., Makiyama, T., Sasaki, K., Yoshida, Y., Wuriyanghai, Y., Chen, J.,Hattori, T., Ohno, S., Kita, T., Horie, M. et al. (2013). Ultrastructural maturation ofhuman-induced pluripotent stem cell-derived cardiomyocytes in a long-termculture. Circ. J. 77, 1307-1314.

Kamath, A. R. and Rai, R. G. (2008). Risk factors for progression of calcific aorticstenosis and potential therapeutic targets. Int. J. Angiol. 17, 63-70.

Kapustin, A. N., Davies, J. D., Reynolds, J. L., McNair, R., Jones, G. T., Sidibe,A., Schurgers, L. J., Skepper, J. N., Proudfoot, D., Mayr, M. et al. (2011).Calcium regulates key components of vascular smooth muscle cell-derived matrixvesicles to enhance mineralization. Circ. Res. 109, e1-e12.

Kay, H. H., Zhu, S. and Tsoi, S. (2007). Hypoxia and lactate production introphoblast cells. Placenta 28, 854-860.

Laemmli, U. K. (1970). Cleavage of structural proteins during the assembly of thehead of bacteriophage T4. Nature 227, 680-685.

Lei, Y., Sinha, A., Nosoudi, N., Grover, A. and Vyavahare, N. (2014).Hydroxyapatite and calcified elastin induce osteoblast-like differentiation in rataortic smooth muscle cells. Exp. Cell Res. 323, 198-208.

Leung, K. S.-Y. and Fong, B. M.-W. (2014). LC-MS/MS in the routine clinicallaboratory: has its time come? Anal. Bioanal. Chem. 406, 2289-2301.

Liberman, M., Bassi, E., Martinatti, M. K., Lario, F. C., Wosniak, J.,Pomerantzeff, P. M. A. and Laurindo, F. R. M. (2008). Oxidant generationpredominates around calcifying foci and enhances progression of aortic valvecalcification. Arterioscler. Thromb. Vasc. Biol. 28, 463-470.

Liu, M., Xu, F., Tao, T., Song, D., Li, D., Li, Y., Guo, Y. and Liu, X. (2016). Molecularmechanisms of stress-induced myocardial injury in a rat model simulatingposttraumatic stress disorder. Psychosom. Med. 78, 888-895.

Lossie, J., Kohncke, C., Mahmoodzadeh, S., Steffen, W., Canepari, M., Maffei,M., Taube, M., Larchevêque, O., Baumert, P., Haase, H. et al. (2014). Molecularmechanism regulating myosin and cardiac functions by ELC. Biochem. Biophys.Res. Commun. 450, 464-469.

Martın-Rojas, T., Gil-Dones, F., Lopez-Almodovar, L. F., Padial, L. R.,Vivanco, F. and Barderas, M. G. (2012). Proteomic profile of human aorticstenosis: insights into the degenerative process. J. Proteome Res. 11,1537-1550.

Martın-Rojas, T., Mourino-Alvarez, L., Alonso-Orgaz, S., Rosello-Lleti, E.,Calvo, E., Lopez-Almodovar, L. F., Rivera, M., Padial, L. R., Lopez, J. A.,de la Cuesta, F. et al. (2016). iTRAQ proteomic analysis of extracellular matrixremodeling in aortic valve disease. Sci. Rep. 5, 17290.

Mayr, M., Grainger, D., Mayr, U., Leroyer, A. S., Leseche, G., Sidibe, A., Herbin,O., Yin, X., Gomes, A., Madhu, B. et al. (2009). Proteomics, metabolomics, andimmunomics on microparticles derived from human atherosclerotic plaques. Circ.Cardiovasc. Genet. 2, 379.

Mehta, D. and Malik, A. B. (2006). Signaling mechanisms regulating endothelialpermeability. Physiol. Rev. 86, 279-367.

Miguet, L., Pacaud, K., Felden, C., Hugel, B., Martinez, M. C., Freyssinet, J.-M.,Herbrecht, R., Potier, N., van Dorsselaer, A. and Mauvieux, L. (2006).Proteomic analysis of malignant lymphocyte membrane microparticles usingdouble ionization coverage optimization. Proteomics 6, 153-171.

Mihaljevic, T., Nowicki, E. R., Rajeswaran, J., Blackstone, E. H., Lagazzi, L.,Thomas, J., Lytle, B. W. and Cosgrove, D. M. (2008). Survival after valvereplacement for aortic stenosis: implications for decision making. J. Thorac.Cardiovasc. Surg. 135, 1270-1279.e12.

Mourino-Alvarez,L., Baldan-Martin,M.,Gonzalez-Calero, L.,Martinez-Laborde,C.,Sastre-Oliva, T., Moreno-Luna, R., Lopez-Almodovar, L. F., Sanchez, P. L.,Fernandez-Aviles, F., Vivanco, F. et al. (2016a). Patients with calcific aorticstenosis exhibit systemic molecular evidence of ischemia, enhancedcoagulation, oxidative stress and impaired cholesterol transport. Int. J. Cardiol.225, 99-106.

Mourino-Alvarez, L., Iloro, I., de la Cuesta, F., Azkargorta, M., Sastre-Oliva, T.,Escobes, I., Lopez-Almodovar, L. F., Sanchez, P. L., Urreta, H.,Fernandez-Aviles, F. et al. (2016b). MALDI-Imaging Mass Spectrometry: astep forward in the anatomopathological characterization of stenotic aortic valvetissue. Sci. Rep. 6, 27106.

New, S. E. P. and Aikawa, E. (2013). Role of extracellular vesicles in de novomineralization: an additional novel mechanism of cardiovascular calcification.Arterioscler. Thromb. Vasc. Biol. 33, 1753-1758.

9

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033423. doi:10.1242/dmm.033423

Disea

seModels&Mechan

isms

Ngo, D. T. M., Stafford, I., Kelly, D. J., Sverdlov, A. L., Wuttke, R. D., Weedon, H.,Nightingale, A. K., Rosenkranz, A. C., Smith, M. D., Chirkov, Y. Y. et al. (2008).Vitamin D2 supplementation induces the development of aortic stenosis in rabbits:interactions with endothelial function and thioredoxin-interacting protein.Eur. J. Pharmacol. 590, 290-296.

Otto, C. M. (2006). Valvular aortic stenosis: disease severity and timing ofintervention. J. Am. Coll. Cardiol. 47, 2141-2151.

Park, S. J., Kim, S. H., Choi, H. S., Rhee, Y. and Lim, S.-K. (2009). Fibroblastgrowth factor 2-induced cytoplasmic asparaginyl-tRNAsynthetase promotessurvival of osteoblasts by regulating anti-apoptotic PI3K/Akt signaling. Bone45, 994-1003.

Perkins, D. N., Pappin, D. J. C., Creasy, D. M. and Cottrell, J. S. (1999).Probability-based protein identification by searching sequence databases usingmass spectrometry data. Electrophoresis 20, 3551-3567.

Peterson, D. B., Sander, T., Kaul, S., Wakim, B. T., Halligan, B., Twigger, S.,Pritchard, K. A., Oldham, K. T. and Ou, J.-S. (2008). Comparative proteomicanalysis of PAI-1 and TNF-alpha-derived endothelial microparticles. Proteomics8, 2430-2446.

Rajamannan, N. M., Evans, F. J., Aikawa, E., Grande-Allen, K. J., Demer, L. L.,Heistad, D. D., Simmons, C. A., Masters, K. S., Mathieu, P., O’Brien, K. D. et al.(2011). Calcific aortic valve disease: not simply a degenerative process: a reviewand agenda for research from the National Heart and Lung and Blood InstituteAortic StenosisWorkingGroup. Executive summary: Calcific aortic valve disease-2011 update. Circulation 124, 1783-1791.

Rossignol, F., Solares, M., Balanza, E., Coudert, J. and Clottes, E. (2003).Expression of lactate dehydrogenase A and B genes in different tissues of ratsadapted to chronic hypobaric hypoxia. J. Cell. Biochem. 89, 67-79.

Ruifrok, A. and Johnston, D. (2001). Quantification of histochemical staining by

color deconvolution. Anal. Quant. Cytol. Histol. 23, 291-299.Seiler, C. and Jenni, R. (1996). Severe aortic stenosis without left ventricular

hypertrophy: prevalence, predictors, and short-term follow up after aortic valve

replacement. Heart 76, 250-255.Shevchenko, A., Wilm, M., Vorm, O. and Mann, M. (1996). Mass spectrometric

sequencing of proteins from silver-stained polyacrylamide gels. Anal. Chem.

68, 850-858.Surinova, S., Schiess, R., Huttenhain, R., Cerciello, F., Wollscheid, B. and

Aebersold, R. (2011). On the development of plasma protein biomarkers.

J. Proteome Res. 10, 5-16.Turk, J. R. and Laughlin, M. H. (2004). Physical activity and atherosclerosis: which

animal model? Can. J. Appl. Physiol. 29, 657-683.Wang, M., Li, W., Chang, G.-Q., Ye, C.-S., Ou, J.-S., Li, X.-X., Liu, Y., Cheang,

T.-Y., Huang, X.-L. and Wang, S.-M. (2011). MicroRNA-21 regulates vascular

smooth muscle cell function via targeting tropomyosin 1 in arteriosclerosis

obliterans of lower extremities. Arterioscler. Thromb. Vasc. Biol. 31, 2044-2053.Weind, K. L., Ellis, C. G. and Boughner, D. R. (2000). The aortic valve blood

supply. J. Heart Valve Dis. 9, 1-7.Weind, K. L., Ellis, C. G. and Boughner, D. R. (2002). Aortic valve cusp vessel

density: relationship with tissue thickness. J. Thorac. Cardiovasc. Surg.

123, 333-340.Yin, M., Loyer, X. and Boulanger, C. M. (2015). Extracellular vesicles as new

pharmacological targets to treat atherosclerosis. Eur. J. Pharmacol. 763, 90-103.

10

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033423. doi:10.1242/dmm.033423

Disea

seModels&Mechan

isms