Identification of common cardiometabolic alterations and deregulated pathways in mouse and pig models of aging

Abstract Aging is the main risk factor for cardiovascular and metabolic diseases, which have become a global concern as the world population ages. These diseases and the aging process are exacerbated in Hutchinson–Gilford progeria syndrome (HGPS or progeria). Here, we evaluated the cardiometabolic disease in animal models of premature and normal aging with the aim of identifying alterations that are shared or specific to each condition. Despite differences in body composition and metabolic markers, prematurely and normally aging mice developed heart failure and similar cardiac electrical abnormalities. High‐throughput proteomics of the hearts of progeric and normally aged mice revealed altered protein oxidation and glycation, as well as dysregulated pathways regulating energy metabolism, proteostasis, gene expression, and cardiac muscle contraction. These results were corroborated in the hearts of progeric pigs, underscoring the translational potential of our findings, which could help in the design of strategies to prevent or slow age‐related cardiometabolic disease.


Supplementary Experimental Procedures
Animal Care. Mouse studies were carried out in 4-month-old progeric Lmna G609G/G609G and Lmna G609G/wt animals (Osorio et al., 2011) and in 4-month-old and 20-month-old wild-type (WT) mice on the C57BL/6 genetic background. When indicated, we used 10-month-old Lmna G609G/wt and WT mice. Unless otherwise specified, experiments were performed with a balanced proportion of male and female mice to avoid gender-biased results. Pig proteomics was performed with castrated ~5-month-old male heterozygous LMNAc.1824C>T Yucatan minipigs (Dorado et al., 2019) or WT Yucatan minipigs. All experimental procedures with mice and pigs were reviewed and authorized by Ethics Committees  in accordance with EU Directive 2010/63EU and Recommendation 2007/526/EC, enforced in Spanish law under Real Decreto 53/2013. Every effort was made to minimize animal suffering.
Heart Rate. Heart rate (HR) was measured with a non-invasive automated tail-cuff device (BP2000, Visitech Systems) in conscious mice to avoid anesthetic interference. Measurements were taken on three consecutive days without applying tail-cuff pressure. To improve accuracy, the first 10 of 20 measurements were discarded. The interquartile range (IQR) was calculated (difference between quartiles 3 and 1), and data points 1.5xIQR below the first quartile (Q1) or 1.5xIQR above the third quartile (Q3) were discarded as outliers.
Electrocardiography. Mice were anesthetized with 1-2% isoflurane, and 4 electrocardiography (ECG) electrodes were inserted subcutaneously into the limbs. ECG recordings were acquired for 2 min at 2 KHz sweep-speed using a MP36R data acquisition workstation (Biopac Systems). ECG data were exported with AcqKnowledge software (Biopac Systems) and automatically analyzed using custom R scripts, which remove artifacts, noise and baseline fluctuations; detect heart beats, peaks and waves; and calculate ECG intervals. The second lead was selected for the study, since the signal was more stable in most experiments, and therefore the identification of waves was more robust using this channel.
PQ interval was measured from the P-wave peak to the beginning of the Q-wave (Qs). The end of the S-wave (J) is not evident in mice because the ST segment is absent in this species; therefore, the QRS complex was calculated from the Qs to the S-wave minimum, and the QT interval was calculated from the Qs to the T-wave peak. The QT90 was defined as the interval between the Qs and the point where the T-wave declines by 90% (T90) from the peak. QTe was also measured, from Qs to the point where the negative component of the T-wave returned to the isoelectric line (T e ). T-wave morphological alterations were quantified by defining the JT gap and the T-wave steepness. The JT gap measures the distance between the J point and the Twave peak, typically coincidental in mice, thus allowing detection of delays in T-wave appearance. T-wave steepness serves as a measure of T-wave flattening because it represents the absolute value of the slope (difference in voltage divided by the difference in time) between the peak of the T-wave and T90.
Echocardiography. Transthoracic echocardiography was performed by an expert operator blinded to genotype using a high-frequency ultrasound system (Vevo 2100, VisualSonics) with a 40 MHz linear probe. Before echocardiography, animal fur was removed with a topical depilatory agent. Two-dimensional (2D) and M-mode echography were performed at a frame rate >230 frames/s and pulsed wave (PW) Doppler was acquired with a pulse repetition frequency of 40 kHz. Mice were lightly anesthetized with 1-2% isoflurane in oxygen through a facial mask, adjusting the isoflurane delivery to maintain the HR at 450 ± 50 bpm. Mice were placed in a supine position on a heating platform, and warmed ultrasound gel was used to maintain normothermia. A base apex ECG was continuously monitored. Images were analyzed off-line using the Vevo 2100 Workstation software (VisualSonics).
For left ventricle (LV) systolic function assessment, parasternal standard 2D and M-mode, long axis (LAX) and short axis (SAX) views were acquired (Moran et al., 2013). LV chamber dimensions were measured from these views, which served to calculate LV ejection fraction (EF) and LV chamber mass (Foppa et al., 2005). Cardiac Output (CO) was calculated by measuring HR, the area of the LV outflow tract diameter (LVOTd) and the velocity time integral (VTI) (Zoghbi & Quinones, 1986). LVOTd was assessed from a 2D LAX view, at the level of the aortic valve annulus, after maximal systolic leaflet separation. VTI was obtained by tracing the LVOT flow signal envelope, which was acquired by aligning the PW Doppler beam with the LV outflow. Mitral Valve (MV) inflow pattern was acquired using PW Doppler echography in the 4-chamber apical view to assess diastolic function. The sample volume was positioned parallel to the blood flow, across the mitral orifice. Measured variables included early and late diastolic velocity peak wave (E and A, respectively), the E/A ratio and isovolumetric relaxation time (IVRT) (Schnelle et al., 2018).
Magnetic Resonance Imaging. Magnetic resonance imaging (MRI) studies were performed at the CNIC Advanced Imaging Unit with a 7-T Agilent/Varian scanner (Agilent Technologies) equipped with a DD2 console. For image acquisition, mice were sedated with 2% isoflurane in oxygen. Ophthalmic gel was applied to prevent retinal drying. Cardiac MRI was performed using an actively shielded 115/60 gradient. A surface coil was used for MRI signal transmission and reception. MRI experiments were conducted by applying an ECG-triggered fast gradient echo cine sequence with the following imaging parameters: repetition time, 125 ms (minimum values depending on HR and number of frames per heart cycle); echo time, 1.25 ms; field of view, 30 x 30 mm; acquisition matrix, 128 x 128; flip angle, 15º; 4 averages; 20 cardiac phases; and 12 0.8 mm slices (depending on heart size) with a 0.2 mm gap.
The examination began with the acquisition of survey chest images in the three orthogonal planes (transverse, coronal and sagittal) to localize the heart and obtain images in twochamber SAX view, 4-chamber view and two-chamber LAX view. By using the 4-chamber and two-chamber LAX views, a series of parallel SAX images were defined (encompassing the entire LV from the base to the apex). The apical-most slice of the multislice set was positioned inferior to the heart, and the basal-most slice was positioned above the LV to ensure total coverage of the ventricles through the cardiac cycle. Body fat images were acquired using a spin echo multi slice (SEMS) sequence with and without fat saturation and with the following parameters: repetition time, 533-884 ms (varies between with and without fat saturation); echo time, 10.44 ms; acquisition matrix, 128 x 128; flip angle, 90º; 1 average; and 35 1 mm slices with a 1 mm gap.
Cine MRI was analyzed using the freely available Segment software (Heiberg et al., 2010). On both ventricles, epicardial and endocardial borders were delineated manually on the SAX of each slide to measure total ventricular volumes and cardiac mass (Hernández-Porras et al., 2014). Cardiac function was assessed by calculating LV and right ventricle (RV) EF, stroke volume (SV) and CO from these measurements. Additionally, the RV to LV end-diastolic volume (EDV) ratio and the Fulton index (RV/LV mass) were determined to analyze ventricle relative size. Whole body images were first bias-field-corrected and segmented using Advances Normalization Tools (Avants et al., 2011). The partial volume estimate corresponding to fat was then masked in order to remove brain tissue from whole body fat composition, and the volume of non-zero voxels was calculated in FSL (Jenkinson et al., 2012). Brain masks were manually drawn in ITK snap (Yushkevich et al., 2006). In cases requiring manual correction of the fat segmentation images, partial volumes were intensity-rescaled and re-segmented. Fat mass was estimated as the product of fat volume and fat density (0.923 g/cm 3 ) (Hill et al., 2007).
Tolerance to Challenge Tests and Blood Analysis. For glucose tolerance tests (GTT), overnight fasted mice received an intraperitoneal injection of glucose (1 g/kg body weight, Sigma-Aldrich). In insulin tolerance tests (ITT) studies, 1 h fasted mice received an intraperitoneal injection of insulin (0.75 IU/kg, Lilly). Blood glucose levels were determined in tail-vein blood at 0, 15, 30, 60, 90 and 120 min post injection with a Contour Next One Smart Meter (Contour Next).
For biochemical analysis, animals were fasted overnight before blood was extracted directly from the mandibular sinus (live mice) or from the heart and renal artery (euthanized mice). Blood samples were collected in plastic tubes and incubated at room temperature for 1 to 2 h to allow clotting. Serum samples were later centrifuged at 1200 g for 15 min and the supernatant was stored at -80 ºC. Biochemical variables were analyzed using a Dimension RxL Max Integrated Chemistry System (Siemens Healthineers).
Sample Preparation for Proteomics. Mice were euthanized in a CO2 chamber, and 0.9% salineperfused hearts were collected, snap-frozen in liquid nitrogen, pulverized with a pestle and mortar and preserved at -80 ºC until use. Protein extracts were obtained by homogenizing tissue with ceramic beads (MagNa Lyser Green Beads apparatus, Roche) in lysis buffer (1.5% SDS, 1 mM EDTA, 50 mM Tris-HCl at pH 8.5) supplemented with 50 mM iodoacetamide (Sigma-Aldrich) to block reduced Cys residues (Martínez-Acedo et al., 2012). Protein concentrations were determined by RC/DC Protein Assay (Bio-Rad Laboratories). We analyzed two samples for each experimental condition, each sample consisting of a pool of 2 hearts, ensuring equal contributions from each biological replicate.
Samples were subjected to tryptic digestion using filter-aided sample preparation (FASP) technology (Expedeon) according to the previously published method (Wiśniewski et al., 2009), which for mouse samples was adapted for oxidized Cys labeling based on the FASILOX method (Bonzon-Kulichenko et al., 2020). Briefly, protein extracts (100 μg) were diluted in urea sample solution (USS, 8 M urea in 100 mM Tris-HCl at pH 8.5) and loaded onto the filters. After centrifugation and a washing step with USS, Cys residues were reduced with 50 mM DTT (GE Healthcare) in preparation buffer (50 mM HEPES, 1 mM EDTA at pH 7.5) for 1 h at room temperature. Next, the samples were centrifuged, washed with preparation buffer and subsequently alkylated with 50 mM S-methyl methanethiosulfonate (MMTS) (Thermo Fisher Scientific) in preparation buffer for 1 h at room temperature. Filters were then washed three times with USS and three times with ABC Buffer (50 mM NH4HCO3 at pH 8.8). Proteins were digested using sequencing grade trypsin (Promega) at a 1:40 trypsin: sample-mass ratio.
Eluted and cleaned-up peptides were subjected to isobaric tag for relative and absolute quantitation (iTRAQ) 8-plex labeling (AB Sciex), combined and desalted on Waters Oasis HLB C18 cartridges (Waters Corp). Labeled mouse peptides were separated into eight fractions using Waters Oasis MCX cartridges (Waters Corp) and graded concentrations of ammonium formate (pH 3.0) (AF3) in acetonitrile (ACN). Briefly, desalted and dried peptides were taken up in 1 ml of 5 mM AF3 containing 25% (v/v) ACN. The MCX cartridge was equilibrated with 1 ml of 1:1 methanol:water, followed by 3 ml of 5 mM AF3 containing 25% (v/v) ACN. After loading the samples, the cartridge was washed with AF3, 25% (v/v) ACN. Bound peptides were eluted with the following solvents: 1) 500 mM AF3, 25% ( Liquid Chromatography Coupled to Tandem Mass Spectrometry. The tryptic peptide mixtures were subjected to nanoscale liquid chromatography (LC) coupled to tandem mass spectrometry (MS/MS). High-resolution analysis of samples was performed on an EASY-nLC 1000 liquid chromatograph (Thermo Fisher Scientific) coupled to either an Orbitrap Fusion mass spectrometer (Thermo Fisher Scientific) or a Q Exactive HF mass spectrometer (Thermo Fisher Scientific). Peptides were suspended in 0.1% formic acid and loaded onto a C18 RP nanoprecolumn (Acclaim PepMap100: internal diameter, 75 μm; particle size, 3 μm; and length, 2 cm; Thermo Fisher Scientific). Peptides were then separated on an analytical C18 nano-column (EASY-Spray column PepMap RSLC C18: internal diameter 75 μm; particle size 3 μm; and length, 50 cm; Thermo Fisher Scientific) in a continuous gradient: 8-27% B for 240 min, 31-100% B for 2 min, 100% B for 7 min, 100-2% B for 2 min and 2% B for 30 min (where A is 0.1% formic acid in HPLC-grade water and B is 90% ACN, 0.1% formic acid in HPLC-grade water).
Spectra were acquired using full ion-scan mode over the 400-1,500 mass-to-charge (m/z) range and 120,000 resolution. The automatic gain control target was set at 2 x 10 5 with a maximum injection time of 50 ms. Data-dependent speed mode acquisition was performed at 5 x 10 4 automatic gain control and 120 ms injection time, with a 1 Da isolation window and 45 s dynamic exclusion. High collision energy dissociation fragmentation was set to 33% normalized collision energy. MS/MS scan resolution was set to 15,000, and the first mass in the fragmentation spectrum range was fixed at 100 m/z. Quantitative Analysis of Proteomic Data. Proteins were identified in the raw files using the SEQUEST HT algorithm integrated in Proteome Discoverer 2.1 (Thermo Fisher Scientific). MS/MS scans were matched against protein databases for the mouse (UniProtKB 2016-07 release) or for the pig (UniProtKB 2018-09 release) supplemented with pig trypsin and human keratin sequences. Database search parameters were as follows: trypsin digestion with maximum 2 missed cleavage allowed, precursor mass tolerance of 800 ppm and a fragment mass tolerance of 0.02 Da. Fixed modifications were the N-terminal and lysine iTRAQ 8-plex modifications, whereas variable modifications were methionine oxidation, Cys carbamidomethylation and Cys methylthiolation. To study AGE posttranslational modifications, several AGEs described in the Unimod database (www.unimod.org) were selected as additional variable modifications: glyoxal-derived hydroimidazolone, methylglyoxal-derived hydroimidazolone, malondialdehyde adduct, dihydroxyimidazolidine, hydroxyphenylglyoxal, phenylglyoxal adduct and bis(hydroxyphenylglyoxal) in arginine; malondialdehyde adduct, 2-ammonio-6-[4-(hydroxymethyl)-3-oxidopyridinium-1-yl]hexanoate and carboxymethylation in lysine; and tryptophan carboxymethylation. SEQUEST results were analyzed by the probability ratio (pRatio) method (Martínez-Bartolomé et al., 2008). Calculation of the false discovery rate (FDR) was based on the results obtained by database searching against the corresponded inverted database using the refined method (Bonzon-Kulichenko et al., 2015;Navarro & Vázquez, 2009). Quantitative information was extracted from the intensity of iTRAQ reporter ions in the low-mass region of the 3MS/MS spectra (García-Marqués et al., 2016).
Comparative analysis of protein abundance changes was conducted with the SanXoT software package (Trevisan-Herraz et al., 2019), which was designed for the statistical analysis of highthroughput, quantitative proteomics experiments and is based on the Weighted Scan-Peptide-Protein (WSPP) statistical model (Navarro et al., 2014). As input, WSPP uses a list of quantifications in the form of log 2 -ratios (each condition versus the mean for control 4-monthold WT mice or WT pigs) with their statistical weights. From these, WSPP generates the standardized forms of the original variables by computing the quantitative values expressed in units of standard deviation around the means (Zq). Proteins quantified with only 1 unique peptide were excluded from further analysis. Within-group and between-group variability was assessed in the samples by fitting unsupervised Machine Learning models (hierarchical clustering and principal component analysis: PCA) and by pairwise correlations between replicates and between conditions. Unless otherwise indicated, correlation tests and subsequent analyses were performed to proteins with a between-replicate absolute mean Z score higher than 1.5 (|mean Zq|>1.5) in at least one condition. Known artifact proteins such as keratin contaminants, trypsin, and several major serum proteins (albumin, globins, serpins and complement factors) were excluded from the datasets after the analysis.
For the study of coordinated protein alterations, we used the Systems Biology Triangle (SBT) model, which estimates functional category averages (Zc) from protein values by performing the protein-to-category integration (García-Marqués et al., 2016). The protein category database was consolidated by including resources from Ingenuity Pathway Analysis (IPA) (Qiagen), CORUM (Helmholtz Zentrum München) and DAVID (Leidos Biomedical Research) (Gene Ontology -GO, KEGG Pathway, BioCarta, InterPro, PIR SuperFamily, SMART). Protein interaction networks were analyzed with STRING (ELIXIR Core Data Resources).
Statistical Analysis and Data Visualization. Experimental conditions were randomized and comparisons in animal studies were made between age-matched groups including male and female mice. No statistical method was used to predetermine sample size. Statistical analysis of the results was performed with custom R scripts that apply multifactorial logistic, linear or generalized linear models to compare groups. Natural-logarithm transformations were considered when the normality condition was not met. Significant differences were considered when p<0.05. The following factors were considered as fixed effects: condition (genotype, age group), sex, weight (MRI) and time (tolerance tests). Sample replicates (HR) and date of measurement were included in the models as random effects. Results were represented in R as beeswarm-boxplots or scatter plots with linear or LOESS curve fitting. Boxes in boxplots indicate quartiles and median; whiskers represent the 1.5xIQR; and non-shaded dots are candidate outliers. The shaded area in scatter plots corresponds to the 95% confidence interval of the LOESS curve. Proteomic networks were interpreted with Cytoscape (The Cytoscape Consortium).
Measurements were performed at least 3 times within a month. B) Cardiac function parameters derived from MRI (n=10 mice per group). Body weight was considered as a covariate in the statistical analysis of heart mass. *, p<0.05; **, p<0.01; ***, p<0.001. Figure 7. Main energy metabolism-related pathways altered in the aging mouse heart proteome. Representative pathways with significant changes were extracted from the systems biology analysis. Each dot represents a protein in the pathway. Zq indicates protein abundance compared with young WT mice.

Supplementary
Supplementary Figure 8. Main cardiac contraction-related pathways altered in the aging mouse heart proteome. Representative pathways with significant changes were extracted from the systems biology analysis. A) Structural and motor proteins. B) Ion channels. C) Signaling pathways. Each dot represents a protein in the pathway. Zq indicates protein abundance compared with young WT mice.