High plasma concentrations of acyl-coenzyme A binding protein (ACBP) predispose to cardiovascular disease: Evidence for a phylogenetically conserved proaging function of ACBP

Abstract

Autophagy defects accelerate aging, while stimulation of autophagy decelerates aging. Acyl-coenzyme A binding protein (ACBP), which is encoded by a diazepam-binding inhibitor (DBI), acts as an extracellular feedback regulator of autophagy. As shown here, knockout of the gene coding for the yeast orthologue of ACBP/DBI (ACB1) improves chronological aging, and this effect is reversed by knockout of essential autophagy genes (ATG5, ATG7) but less so by knockout of an essential mitophagy gene (ATG32). In humans, ACBP/DBI levels independently correlate with body mass index (BMI) as well as with chronological age. In still-healthy individuals, we find that high ACBP/DBI levels correlate with future cardiovascular events (such as heart surgery, myocardial infarction, and stroke), an association that is independent of BMI and chronological age, suggesting that ACBP/DBI is indeed a biomarker of "biological" aging. Concurringly, ACBP/DBI plasma concentrations correlate with established cardiovascular risk factors (fasting glucose levels, systolic blood pressure, total free cholesterol, triglycerides), but are inversely correlated with atheroprotective high-density lipoprotein (HDL). In mice, neutralization of ACBP/DBI through a monoclonal antibody attenuates anthracycline-induced cardiotoxicity, which is a model of accelerated heart aging. In conclusion, plasma elevation of ACBP/DBI constitutes a novel biomarker of chronological aging and facets of biological aging with a prognostic value in cardiovascular disease.

Trial registration: ClinicalTrials.gov NCT04879316.

Keywords: aging; autophagy; cancer; cardiovascular diseases; diazepam-binding protein; metabolism.

FIGURE 1

Acb1 (homolog of ACBP) and Ste3—deficient diploid Saccharomyces cerevisiae strains show a (macro)autophagy‐dependent increase in chronological lifespan. (a and b) autophagy immunoblotting analysis of protein extracts from wild type (WT) and ∆acb1 cells expressing a chromosomal GFP‐Atg8 fusion protein. Blots were probed with antibodies against GFP to detect GFP‐Atg8 and free GFP, which is indicative of autophagic flux, and against GAPDH as loading control, revealing a significant increase of autophagic flux in ∆acb1 mutant strains. Representative results (a) and densitometric quantification (b) at 1, 2, and 3 days are shown (n = 5–7). (c and d) autophagy immunoblotting analysis of protein extracts from wild type (WT) and ∆ste3 (coding for membrane receptor that couples factor a pheromone binding to a MAP kinase cascade) cells expressing a chromosomal GFP‐Atg8 fusion protein. Blots were probed with antibodies against GFP to detect GFP‐Atg8 and free GFP, which is indicative of autophagic flux, and against GAPDH as the loading control. Representative results (c) and densitometric quantification (d) at 1, 2, and 3 days are shown (n = 3). Quantitative results are reported as means ± SEM. (e) Chronological aging experiments of wild type and STE3 and ACB1 single gene deletion mutants (∆acb1 and ∆ste3) (n = 4; p values obtained by 2‐way ANOVA: Wt vs. ∆acb1 p = 5 E‐10; wt vs. ∆ste3 p = 0.0002; ∆acb1 vs. ∆ste3 p < 1 E‐11). (f–h) Chronological aging experiments of ACB1 and autophagy‐incompetent ATG single gene deletion mutants (∆acb1, ∆atg5 (f), ∆atg7 (g), and ∆atg32 (h)) as well as double‐mutants thereof (∆acb1/∆atg5, ∆acb1/∆atg7 and ∆acb1/∆atg32). Prolonged CLS by Acb1 deficiency depended on the macro‐autophagic core machinery (Atg5 and Atg7) but was largely independent of mitophagy (Atg32). Strains were grown in batch cultures, dead cells were identified via flow cytometry analysis following propidium iodide (pi) staining and survival was normalized to day one (n = 4 to 6; p values obtained by 2‐way ANOVA: Wt vs ∆acb1 p = 7.9 E‐08; wt vs ∆atg5 p = 1.3 E‐08; wt vs ∆acb1/atg5 p = 0.766; ∆acb1 vs ∆acb1/atg5 p = 1.4 E‐08; ∆atg5 vs ∆acb1/atg5 p = 2.5 E‐09; wt vs ∆atg7 p = 1.5 E‐08; wt vs ∆acb1/atg7 p = 0.004; ∆acb1 vs ∆acb1/atg7 p = 2 E‐10; ∆atg7 vs ∆acb1/atg7 p = 8.7 E‐07; wt vs ∆atg32 p = 0.082; wt vs ∆acb1/atg32 p = 3.5 E‐07; ∆acb1 vs ∆acb1/atg32 p = 1.8 E‐05; ∆atg32 vs ∆acb1/atg32 p = 1.7 E‐09). Statistical comparisons were performed by 2‐way ANOVA (b, d–h) and the corresponding p values are reported on the plots or in their legend.

FIGURE 2

Patients who will develop age‐related pathologies have increased ACBP/DBI levels in the exploration cohort (DESIR 1) and the validation cohort (DESIR 2). (a) The exploration cohort was derived from a weight gain and loss cohort (DESIR 1), of which were drawn the patients who developed cancer or cardiovascular disease during the 9 years follow‐up period and controls matched by age and BMI. (b) Comparison of ACBP/DBI plasma levels (ng/ml) from patients with future cancer or cardiovascular disease versus control individuals from the exploration cohort shows increased levels in patients with future age‐associated diseases. Statistical comparison was performed by one‐sided Student's t‐test and the corresponding p‐value is reported. (c) The validation cohort (DESIR 2) was selected with the same method as the exploration cohort but among the whole DESIR study patients (n = 5212), except for the ones already included in DESIR 1. The final number of patients included those who developed cancer or cardiovascular disease (n = 265) and healthy controls (n = 834). ACBP/DBI plasma levels (ng/ml) of the DESIR 2 patients with the future diagnosis of cancer (d), cardiovascular disease (e), or both (f) as well as their respective controls. Control patients are those who developed neither of these two pathologies during the 9‐years follow‐up period, matched by age and body mass index (BMI) to each case (3:1, respectively). Statistical comparisons were performed by one‐sided Student's t‐test and the corresponding p‐value is reported on the boxplots.

FIGURE 3

The cardiovascular disease‐related increase in ACBP is above that induced by age or body mass index (BMI). The expected positive correlations between ACBP/DBI and age (a), or BMI (b) are observed in the DESIR 2 cohort at the whole‐cohort level (n = 1098) and in controls who developed neither of the two analyzed age‐related pathologies (n = 833). The correlation with age (a) is not significant in the categories of patients who developed cancer (n = 169), cardiovascular disease (n = 96), or one of these two complications (n = 265). The correlation with BMI (b) is maintained in all groups of patients except for the ones who developed cardiovascular disease (n = 94). Pearson's correlation coefficient (R) with their p‐value and the number of samples available (n) are shown in the legend of each panel.

FIGURE 4

A meta‐analysis of publicly available cohorts confirms the weakened correlation between ACBP/DBI plasma levels and age or body mass index (BMI) in the presence of age‐related pathologies. ACBP/DBI levels in the plasma (ng/ml) of patients diagnosed with stage 3 or stage 4 nonsmall cell lung cancer from Cochin hospital were compared by scatter plot with linear regression line with their age (a) or and body mass index (kg/m²) (b). Pearson's correlation coefficient (R) and the associated p values are presented in the legend of each panel, showing a weakening of the correlation of ACBP/DBI with age and a reversion of this correlation to an anticorrelation with BMI. The meta‐analysis of all publicly available cohorts was performed to discriminate between people with cancer or cardiovascular disease (CVD), as well as in patients with neither of these pathologies. Each Pearson's correlation coefficient (COR) is represented with its 95% confidence interval. The size of the square around the COR value is proportional to the sample size of the study. The pooled correlation coefficient was calculated using a random effect model and presented with its 95% confidence interval.

FIGURE 5

ACBP/DBI correlates with conventional cardiovascular risk factors.The correlation of ACBP/DBI with clinical blood parameters was performed in all patients from DESIR 2. Total cholesterol (a), triglycerides (b), systolic blood pressure (c), glycemia (d) correlate positively with the plasma concentration of ACBP/DBI, while and HDL (e) and glomerular filtration rate (f) are inversely correlated with plasma ACBP/DBI. Statistical analysis was performed by linear regression and Pearson's correlation coefficient (R) with the associated p‐value and the number of samples available (n) are shown in the legend of each panel.

FIGURE 6

ACBP neutralization attenuates anthracyline‐induced cardiotoxicity. (a) Doxorubicin (DOX) was administered to C57Bl/6J female mice, which were also treated with an ACBP‐neutralizing antibody (anti‐ACBP) or mouse isotype IgG (CTRL) for the indicated period, before undergoing echocardiography‐based assessment of the heart. (b) Representative echocardiography‐derived left ventricular (LV) M‐mode tracings. (c) LV ejection fraction (LVEF). (d) LV end‐diastolic volume normalized to body surface area (LVEDV_i). (e) LV mass index (LVmass_i), calculated as the ratio between LVmass and body surface area. (f) Tibia length‐normalized lung weight (LW/TL) (g) representative immunoblot of hearts (one heart per lane) from mice of the four treatment groups. (h–k) quantification and statistical analyses of immunoblots from all mice included in the experiment. N = 3–10 mice per group. p values in (c–f, h–k) represent pairwise comparisons between anti‐ACBP‐treated mice and their respective isotype (iso)‐treated controls using simple main effects of a factorial ANOVA. Bars and error bars show means and SEM, respectively, with individual data points superimposed. Echo, Echocardiography; GAPDH, glyceraldehyde‐3‐phosphate dehydrogenase; LC3, Microtubule‐associated proteins 1A/1B light chain 3B; p16, Cyclin‐dependent kinase inhibitor 2A (CDKN2A); p62, Sequestosome‐1 (SQSTM1).