Effects of aging and type 2 diabetes on cardiac structure and function: Underlying mechanisms

We characterized long-term changes in cardiac structure and function in a high-fat diet/streptozotocin mouse model of aging and type 2 diabetes mellitus (T2D) and examined how the intersection of both conditions alters plasma metabolomics. We also evaluated the possible roles played by oxidative stress, arginase activity and pro- inflammatory cytokines. C57BL/6 male mice (13-month-old) were used. Control animals (n = 13) were fed regular chow for 10 months (aged group). T2D animals (n = 25) were provided a single injection of strepto- zotocin and fed a high fat diet for 10 months. In select endpoints, young animals were used for comparison. To monitor changes in left ventricular (LV) structure and function, echocardiography was used. At the terminal study (23 months), blood was collected and hearts processed for biochemical or histological analysis. Echo yielded diminished diastolic function with aging and T2D. LV fractional shortening and ejection fraction decreased with T2D by 16 months peaking at 23 months. Western blots noted increases in fibronectin and type I collagen with aging/T2D and greater levels with T2D in α -smooth muscle actin. Increases in plasma and/or myocardial protein carbonyls, arginase activity and pro-inflammatory cytokines occurred with aging and T2D. Untargeted metabolomics and cheminformatics revealed differences in the plasma metabolome of T2D vs. aged mice while select classes of lipid metabolites linked to insulin resistance, were dysregulated. We thus, document changes in LV structure and function with aging that in select endpoints, are accentuated with T2D and link them to increases in OS, arginase activity and pro-inflammatory cytokines.


Introduction
The prevalence of type 2 diabetes mellitus (T2D) is a growing health concern worldwide as it is estimated that by 2025 there will be 570+ million patients (Lin et al., 2020). The rise in T2D cases is largely attributed to increases in the incidence of obesity and the aging population. Evidence demonstrates that T2D can alter cardiac structure and function, even in the absence of hypertension or coronary artery disease (Boudina and Abel, 2007). This condition is termed diabetic cardiomyopathy (DC). In humans, DC is characterized at early stages by diastolic left ventricular (LV) dysfunction and later progressing to systolic dysfunction, which is accompanied by chamber hypertrophy and fibrosis (Maya and Villarreal, 2010). DC can ultimately progress to congestive heart failure (Boudina and Abel, 2007). Progression typically occurs over many years, if not decades, whereby aging can further contribute to detrimental changes in cardiac structure and function. However, how aging contributes to the evolution of DC is poorly understood and animal models are lacking. It is well established that in the aging population, increases in pro-inflammatory cytokine blood levels occur which are associated with enhanced production of reactive oxygen species (ROS) and the ensuing oxidative stress (OS) (Rea et al., 2018). Significant increases have been reported for cytokines such as tumor necrosis factor (TNF)-α, interleukin (IL)-1β, interferon (IFN)-γ and transforming growth factor (TGF)-β1. These cytokines are recognized for their capacity to activate pro-inflammatory and pro-fibrotic pathways in tissues and organs including the heart (Frangogiannis, 2019). Additional mechanisms that can contribute to tissue fibrosis in the setting of aging and T2D include, increases in arginase activity which shifts arginine utilization towards proline synthesis, an essential component of collagen (Caldwell et al., 2018). The extent to which arginase and cytokine levels increase with aging and T2D is poorly understood.
The development of metabolomic profiling technology has allowed us to gather insight into changes that take place in plasma and/or tissue samples that follow the development of T2D (Chen and Gerszten, 2020). However, plasma metabolomic profiling studies that examine the changes that take place in the setting of an aged model of T2D are currently lacking.
Therefore, the purpose of this study was to characterize changes in LV structure and function in a high-fat diet/streptozotocin (HFD/STZ) mouse model of aging and T2D and how the intersection of both conditions alters plasma metabolomics. Furthermore, we wanted to assess the possible underlying roles played by OS, enhanced arginase activity (as a possible source of increased collagens) and pro-inflammatory cytokines.

Methods
Please see the Supplementary section for a detailed description of methods used. Table 1 and Supplementary Fig. 2 (S2A) summarize recorded animal weights. Average body weight of aged animals demonstrated essentially, no changes over time. For T2D animals, significant increases were noted since the first month of HFD feeding which progressed for 4 months, followed by a stabilization for the following 6 months. At the terminal endpoint, T2D body weight averaged ~57 g vs. ~34 g of aged animals. Fig. S2B plots mean glucose values over time (as per glucose tolerance test; GTT). Aged animals yielded values within normal ranges (baseline glucose of ~75 mg/dL) whereas T2D animals, yielded significantly higher values at baseline (~140 mg/dL) and over time. The GTT area under the curve was significantly higher in T2D mice (33,480 arbitrary units) vs. aged mice (20,415 units) (p < 0.05 by t-test). Thus, as expected, the combined use of a low dose STZ and HFD yielded greater increases in body weight and altered GTT values reflecting a T2D like status vs. aged animals.

Effects of aging and T2D on cardiac structure and function
Echocardiography results are summarized in Table 1 and select endpoints illustrated in Fig. 1.
As shown in Fig. 1A, results reveal a time-dependent effect on estimated LV mass in aged and T2D animals which was significantly greater at 23 months with T2D. In concordance with mass data, values reported in Table 1 for diastolic and in select systolic LV dimensions indicate that aged animals increased chamber size in a moderate but significant manner, whereas T2D hearts increased to an even greater degree. Parameters associated with diastolic function (E ′ /A ′ and deceleration time; DT) yielded at 23 vs. 13 months significant increases or trends in aged animals (p = 0.072 and 0.028 respectively). In the T2D group, E ′ /A ′ and DT increased significantly. Nonetheless, the comparison of E ′ /A ′ or DT values at the 23-month time point between aged and T2D groups were not different (Table 1). Fig. 1C reports on fractional shortening which significantly decreased as early as 16 months of age in T2D animals yielding a value of 26 % at 23 months vs. aged animals (37 %). Fig. 1D reports on serial decreases in ejection fraction in T2D animals yielding a value of 51 % at 22 months which, was significantly lower vs. aged animals (67 %). Thus, in aged T2D animals adverse, time dependent changes occurred in endpoints associated with LV size, diastolic and systolic function. Fig. 2A illustrates representative Western blot images obtained from the LV of young, aged and T2D for type I, III collagens, α-smooth muscle actin (SMA) and fibronectin. To account for loading differences blots were probed with S6RP and results normalized. Fig. 2B reports on relative protein changes noted for aged and T2D animals vs. young (normalized to = 100 %). Significant increases vs. young occurred in aged and T2D animals in all four proteins. Further significant increases in T2D vs. aged occurred in α-SMA. Representative images obtained from immunostaining of LV sections with α-SMA and fibronectin are shown in Fig. S3A. As can be observed, increases were noted for both proteins in aged and T2D animals vs. young. As shown in Fig. S3B, trichrome stained myocardial sections also denoted increases in perivascular fibrosis with aging and T2D. The quantitative analysis of perivascular fibrosis yielded 1.48 ± 0.38 % collagen area fraction in young, 3.21 ± 0.6 % in aged (p = 0.01 vs. young) and 4.23 ± 0.53 (p = 0.0004 vs. young). Thus, with aging significant increases in markers associated with cellular (SMA) and extracellular matrix content increased and these changes trended towards higher levels with T2D.

Effects of aging and T2D on protein carbonylation, cytokines and arginase activity
Values of plasma protein carbonyl levels ( Fig. 3A) detected in young animal samples were 4 ± 1 mmol/ml and increased significantly in aged (10 ± 1 nmol/ml) and further increased in T2D animals (13 ± 1 nmol/ ml). Plasma levels of the pro-inflammatory cytokines TNF-α, IL-1β, IFN-γ and TGF-β1 (panels B-E) significantly increased in aged and T2D animals vs. young. IL-1β, IFN-γ and TGF-β1 further increased in T2D vs. aged animals. Myocardial TGF-β1 levels also increased with aging and T2D (panel F) as well as myocardial and plasma arginase activity ( Fig. S4A and B respectively). Thus, T2D can exacerbate select drivers of aging associated remodeling.

Effects of aging and T2D on untargeted plasma metabolomics
Liquid chromatography-mass spectrometry (LC-MS) 2 data extraction and filtering by the XCMS software retrieved 1726 aligned features or potential metabolites among aged, T2D mice and quality control (QC) datasets. To provide a metabolome overview, we submitted the list of XCMS-extracted features data to Metaboanalyst 4.0 for principal component analysis (PCA). The tight clustering of QC samples (Fig. 4A) demonstrated a reproducible LC-MS 2 methodology. The first and second principal components clearly separated aged and T2D mice, hinting at group metabolic differences. We also utilized heatmap representation and hierarchical clustering analysis, which allowed us to observe specific patterns between datasets using the top-50 most significant features based on the ANOVA test (Fig. S5). Two principal clusters were noted that contrast feature abundance (see box colors) of those 50 metabolites among aged and T2D mice and the abundance homogeneity across QC samples. To identify or annotate the aligned features among groups, we analyzed the LC-MS 2 datasets using Global Natural Products Social Molecular Networking web platform (GNPS), which enables online annotation (Metabolomics Standard Initiative; MSI level 2) by public spectral library matching and in silico (Network Annotation Propagation; NAP and MS2LDA software) metabolite structural predictions (MSI level 3). A molecular network with 403 features (containing MS 2 ) was created wherein each node represents a unique molecule and edges connecting nodes indicate structural similarity (according to a cosine score) (Fig. 4B). Within the network, 66 (62 unique metabolites, Table 1 Body weight, heart rate and echocardiographic assessment of systolic and diastolic function of aged and T2D C57BL/6 38 male mice. Values are mean ± SEM. BW (body weight, g); HR (heart rate, bpm); E  Table S3). According to Classyfire's ontology, the top-10 most representative chemical classes are shown in Fig. 4B. Lipid-related metabolites were predominant within the molecular network. A total of 198 features were differentially abundant between aged and T2D mice, based on foldchange and t-test analysis (p < 0.05) (Fig. 4C). Of the 62 automatically identified metabolites (Supplementary Table S2), only 5 were found differentially abundant among groups based on fold-change and ttest analysis (p < 0.05). Metabolite substructure recognition by MS2LDA further identified 78 metabolites containing a phosphocholine head group (including the three GNPS annotated phospholipids) and of those,   12 were dysregulated in T2D vs. aged mice (Figs. 4D and S6). Specifically, 6 presented with increased abundance (fold change range 2.9 to 7.2), while 6 presented with reduced abundance (fold change range − 2 to − 10) in the T2D group. Another set of metabolites affected in T2D were acylcarnitines. Stearoyl (C18) and palmitoyl (C16) carnitine were putatively annotated by GNPS spectral matching and supported by MS2LDA substructure recognition analysis (Fig. 4E). A third acylcarnitine metabolite was manually annotated (aided by MS2LDA) as oleoyl carnitine (C18:1), but its plasma levels were not affected by T2D.
To overcome the limited annotation of metabolites by spectral matching, we utilized CANOPUS, a computational method that predicts the chemical class of metabolites (MSI level 3) using MS 2 data, independently of spectral libraries or databases. CANOPUS allowed us to provide a global overview of the chemical classes linked to the metabolites (<860 Da) affected in T2D mice. Fatty acyls, glycerophospholipids, carboxylic acids and derivatives were the top classes assigned to the dysregulated metabolites in T2D mice (Fig. 4F). Thus, untargeted metabolomics and cheminformatics reveal distinct differences in the T2D plasma metabolome vs. that of aged mice while select classes of lipid metabolites associated to insulin resistance were dysregulated.

Discussion
Published rodent studies on DC commonly examine disease evolution for <6 months. Common nongenetic models of T2D use rodents subjected to a HFD developing central adiposity, insulin resistance and hyperglycemia (Hsueh et al., 2007). A HFD can be combined with STZ to decrease β-cell insulin secretory capacity leading to a more severe T2D (Hsueh et al., 2007). Another common T2D model is the leptin deficient (i.e., db/db) mouse (Hsueh et al., 2007). In humans, the development of a DC typically takes place over decades (Sattar et al., 2019). Thus, animal models that may better replicate T2D in humans would need to incorporate an aging component. Interestingly, db/db mice possess a life span ~11 months or ~40 human years (Sataranatarajan et al., 2016). With the intent of better replicating the human condition we thus, selected 13-month-old mice or ~45-year-old human that would then evolve for an additional 10 months yielding a human age of ~65 years. Using 2-month-old C57BL/6 mice subjected to HFD/STZ, we previously reported changes in cardiac structure and function that occurred over 6 months (Fricovsky et al., 2012). Diastolic dysfunction developed at the 2-month time point of their T2D. Decreases in fractional shortening were detected at 4 and 6-months. LV chamber size increased modestly but no cardiac hypertrophy or fibrosis was detected. In this study, loss of diastolic function occurred in both groups at the 23-months. This result differs from our previous study and may relate to the pathophysiological impact that early stage T2D using 2-month-old animals has on diastolic function vs. those exposed to T2D at a later age (13-month-old). In aged T2D animals, loss of contraction occurred early and progressed until 23 months. While no recordings of heart weight were made the 30 % increase in echo-based LV mass does not mirror the 60 % gain in body weight. Results also document an excess deposition with aging of extracellular matrix proteins fibronectin and collagen types I and III while those of α-SMA increased further with T2D which, occurred in a diffuse and localized (i.e., perivascular) manner.
Arginase, is an enzyme in the urea cycle that competes with eNOS for L-arginine to produce urea and L-ornithine (Caldwell et al., 2018). Lornithine is metabolized by ornithine aminotransferase and pyrroline-5carboxylate reductase to generate proline for collagen synthesis. Arginase upregulation is linked to endothelial dysfunction by limiting Larginine to eNOS (Garate-Carrillo et al., 2020) resulting in decreased NO synthesis as seen with aging and T2D (Caldwell et al., 2018). We reported increased arginase activity in aged endothelial cells and in ischemic myocardium (Garate-Carrillo et al., 2020;Ortiz-Vilchis et al., 2018). To explore the potential contribution of increased arginase activity in the development of fibrosis we quantified enzyme activity in plasma and myocardium. Activity rises with aging and tended to further increase with T2D mirroring, clinical studies (Romero et al., 2008).
Excess production of ROS with aging is well documented and is associated with mitochondrial dysfunction and lower ATP levels events, that are further enhanced with T2D (Rea et al., 2018). Using 26-monthold mice, we previously reported on increases in brain, kidney, heart, and skeletal muscle protein carbonyls (Moreno-Ulloa et al., 2015). In the current study, we utilized as a surrogate of OS plasma protein carbonyls. Results indicate that with aging a doubling of carbonylation levels occurred that was enhanced by ~20 % with T2D. Enhanced ROS production is linked to increases in inflammatory cytokines which contribute to cardiac remodeling. Increases in blood TNF-α levels have been reported in mouse models of accelerated (Miro et al., 2017) or natural aging (Yamamoto et al., 2002) and mirror those reported in humans (Bruunsgaard et al., 2000). Plasma TNF-α levels also increase in T2D mouse models (Yamamoto et al., 2002;Gao et al., 2007) and in T2D patients (Nilsson et al., 1998;Pickup et al., 2000). Combined effects of aging and T2D on TNF-α serum levels were reported in the db/db mouse (Zhang et al., 2008). In humans, TNF-α levels in normal elderly were accentuated with T2D (Pedersen et al., 2003). Preclinical studies have also reported increases in IFN-γ levels with aging (Cayetanot et al., 2009;Rodriguez et al., 2007) and T2D (Hazman and Ovali, 2015). In humans, IFN-γ levels increase with T2D (Li et al., 2020). Interestingly, with IL-1β, studies are controversial (Barrientos et al., 2009). We detected increases in all cytokines with aging with further elevations in IFN-γ and IL-1β with T2D.
TGF-β1 is a key fibrosis orchestrator. Increases in plasma TGF-β1 are reported in aged rats (Chacar et al., 2019) and in mice (Cai et al., 2012). Aging associated increases in TGF-β1 levels are also reported in humans (Carrieri et al., 2004;Forsey et al., 2003). Yadav et al., reported increases in TGF-β1 levels in obese animals and in overweight or obese humans (Yadav et al., 2011). Myocardial TGF-β1 levels also increase in obese rabbits (Carroll and Tyagi, 2005). In patients, serum TGF-β levels increase with T2D (Azar et al., 2000). In control and in diabetic rats, plasma TGF-β1 levels increased with aging however, in older diabetic rats TGF-β1 increased further (Hosomi et al., 2002). Here, we report aging associated increases in TGF-β1 plasma and myocardial levels. For TGF-β1 plasma, levels for aged animals tripled vs. young and levels further increased with T2D.
Untargeted plasma metabolomics allowed us to determine global and specific differences among cohorts. We employed complementary dereplication strategies to annotate metabolites at the structure and substructure level (Ramirez-Sanchez et al., 2021). Of the different metabolites identified, we noted that lipids were principally dysregulated with T2D vs. aged mice. Our results are similar to those of Kim et al., who revealed that fatty acids, glycerophosphocholines and acylcarnitines were altered in mice after a 3 month HFD (Kim et al., 2011). Such metabolites are thought to be involved in the development of obesity and T2D in animals (Horakova et al., 2016) and humans (Mihalik et al., 2010;Adams et al., 2009). Interestingly, long-chain acylcarnitines (C16 and C18) were shown to serve as a biomarker of propensity to obesity and partially, for obesity-associated insulin resistance in HFD mice (Horakova et al., 2016). Long-chain acylcarnitines are also suggested as early T2D biomarkers in humans (Sun et al., 2016). A reported positive correlation between total sum of plasma acylcarnitines and glycated hemoglobin suggests that this type of metabolites may also reflect glucose metabolism in T2D patients (Adams et al., 2009). Acylcarnitines are produced to import long-chain fatty acids into mitochondria for energy production via fatty acid β-oxidation and increased levels are associated with altered fatty acid β-oxidation (Adams et al., 2009). Interestingly, the abnormal accumulation of stearoylcarnitine (C18) in the pancreas of prediabetic and overtly diabetic mice correlates with β cell dysfunction. Moreover, in-vitro studies have reported that treatment with long-chain (C14, C16, C18) acylcarnitines of skeletal muscle myotubes triggers OS and inflammation (McCoin et al., 2015). Long-term studies are thus, needed to determine if the elevated levels of select lipids precede hyperglycemia, are the consequence of impaired oxidation or both (vicious cycle).
Another group of T2D dysregulated metabolites was a panel of glycerophosphocholine-containing lipids. Our results are similar to those of Kim et al.,who showed that LysoPC (18:3) and LysoPC (14:0) were respectively up-modulated and down-modulated in HFD mouse plasma (Kim et al., 2011). We also identified various LysoPC lipids varying in length but were not altered with T2D. Therefore, results suggest that an HFD dysregulates glycerophospholipids' metabolism. Interestingly, some studies have reported critical roles of select lipid metabolites on insulin resistance (van der Veen et al., 2019) and inflammation (Murugesan et al., 2003).

Limitations
There are well recognized limitations in the use of small animal models of diabetes as they do not fully recapitulate human disease. However, they provide advantages as per their "accelerated" aging and also useful introspect for changes in critical endpoints. The animal model of HFD/STZ T2D is widely used but, has been controversial. The study was only performed in male animals as per their established and reliable capacity to be induced into a T2D like state using a HFD and low dose of STZ where reliable alterations in glucose tolerance can be triggered. The study did not include a female counterpart of T2D as it is well established that female mice are more resistant to diabetes and obesity induction using diet and/or drugs (Kennard et al., 2021). Female mice are resistant to the STZ doses used in males. Increasing the dose of STZ can elicit diabetes in females however, at a much greater cost of as per hepatic and renal toxicity (Kennard et al., 2021).

Conclusions
Using a mouse model we provide for the first time, an extended window into the development of DC which, may better reflect the prolonged evolution seen in humans and the challenges or opportunities that may be encountered to alter the course of the disease with targeted therapies.

Funding
This work was supported by the Veterans Administration Merit Award (1I01BX003230), Department of Defense PR150090 and National Institutes of Health DK98717, AG47326 to F.V. Plasma metabolomics was funded in part by the National Council of Science and Technology, Mexico (CONACyT), project 300437 to A.M.U.

Declaration of competing interest
Dr. Villarreal is a co-founder of Epirium Bio, Inc. and stockholder (Dr. Ceballos).