With serial assessment of left ventricular function in a cohort of mice, we have described the age at which mice develop systolic dysfunction, and focused our tissue analysis on this timepoint. We have demonstrated that aging male C57/Bl6 mice: 1) develop contractile dysfunction at 18 months of age, and this is associated with structural changes of increased fibrosis and cardiomyocyte hypertrophy; 2) have a different gene expression profile in the left ventricular myocardium from young mice with upregulation of immune/inflammatory genes; and 3) have an increase in activated caspase-3 in cardiomyocytes, with a parallel upregulation of anti-apoptotic factors.
The very existence of an age-related cardiomyopathy is controversial. Some authors have found that with age the murine heart develops contractile dysfunction [4
], whereas other have not [7
]. Still other groups have shown that aging is associated with normal baseline cardiac function, but reduced cardiac response to inotropic stimuli [11
]. However, even among those who have demonstrated an age-related cardiomyopathy, the age at which it occurs is still a matter of some debate. In trying to make sense of such varied and contradictory literature, several issues must be taken into account. Firstly, there are differences in strain and age of the mice in each of these studies, and there will likely be differences in the incidence of heart failure across strains and ages. There may even be differences within strains in the age at which age-related cardiomyopathy occurs, due to different living/cage conditions at various institutions, and different feeding parameters. Secondly, the absence of heart failure at one age in an experiment does not mean it will not occur at a later time in that strain of mouse. Most studies do not serially evaluate cardiac structure and function in the same cohort over time, as we have. We feel that our method of serial echocardiography demonstrating deterioration of function in previously normal hearts is a more robust proof of the existence of age-related cardiomyopathy than simply comparing groups of different animals at different ages. Finally, there are many ways to evaluate cardiac function, including echocardiography, LV pressure measurements and ex vivo
whole heart perfused preparations. There are many choices for anesthetic agents, and all have some effect on cardiorespiratory function. Thus, the different loading conditions during the different techniques used, and the type of anesthetic agent employed, can have significant effects on the results of the experiments and make comparisons between studies problematic. Thus, we kept our anesthetic agents and doses the same at each study, used the same echocardiographic machine and echocardiographer, in order to keep the variables to a minimum. This approach strengthens our findings, and we conclude that our results confirm the presence of a murine age-related cardiomyopathy, and describe its age of onset in our laboratory.
Our findings confirm previous studies showing LV fibrosis with increasing age. Interestingly, we have extended these findings to show that this is not associated with upregulation of any known pro-fibrotic genes, or any known fibrotic gene pathways. Although increased cardiac fibrosis has been shown in other disease states in the absence of upregulation of fibrosis-related pathways [18
], to our knowledge this is the first description of age-associated increase in cardiac fibrosis without upregulation of any known pro-fibrotic pathway. TGFb is thought to be a crucial mediator of age-related cardiac fibrosis [20
]. We showed no change in expression between young and aging hearts in TGFb subtypes 1, 2 and 3, nor any change in their receptor expression. We then used gene set enrichment analysis to determine whether any known fibrosis pathways were involved in our age-related cardiac fibrosis. GSEA is a publicly available resource at www.broadinstitute.org/gsea/
that allows users to analyze their microarray data against known pathways (that are also available on their website). This allows for detection of smaller deviations between groups of known clusters of genes associated with a biological function. Gene expression changes that are too small to be detected using standard multiple comparison testing can be detected with GSEA [16
]. The lack of any gene changes or pathway differences suggests that the fibrosis is either due to novel fibrosis pathways, to non-transcriptionally mediated pathways, or to reduction in turnover in the extracellular matrix. We did not see down regulation of collagen turnover pathways such as the tissue inhibitors of matrix metalloproteinases (TIMPS) at the RNA level, although increased clearance of these or other collagenases could potentially account for these findings. Willems et al [21
] also demonstrated increased collagen content at 18 months of age in the mouse heart. In contrast to our study, they showed no difference in left ventricular function at the same age despite the increased collagen content. Differences in experimental technique may have accounted for these different findings, as they employed ex-vivo Langendorrf preparation to assess ventricular function, whereas we used in vivo echocardiography. The established fibrosis in our study right at the onset of the LV dysfunction suggests that the fibrosis had been occurring for some time before the development of LV dysfunction, and thus is more likely to be involved in the pathogenesis of the cardiomyopathy than reactive to it. The findings of WIllems et al are consistent with this hypothesis. It is conceivable that they examined LV function after the onset of fibrosis but before the onset of LV dysfunction. In combination with fibrosis, also we found significant cardiomyocyte hypertrophy. Although the phenotype is similar to hypertensive heart disease, this was not due to elevated blood pressure and there was no increase in the known pathways for LV hypertrophy on microarray analysis. The hypertrophy seen with age may therefore occur via different gene pathways from hypertrophy that occurs in known disease states.
Our gene expression data revealed an increase in several gene ontology terms associated with immunity and/or inflammation. Mast cells are associated with myocardial fibrosis [22
], and the number of mast cells in the heart has been shown to correlate with the amount of myocardial fibrosis [23
]. To explore this link between mast cell number and fibrosis in age-related cardiomyopathy, we examined the number of mast cells in the hearts of young and aging animals but found no difference. Macrophage infiltration in the heart is also associated with pathological fibrosis in experimental models [24
]. We therefore examined our heart tissues for macrophages and found there were very few macrophages at all, with no difference in the numbers between young and aging mice hearts. We detected clusters of lymphocyte-like cells in the peri-vascular regions of aging hearts. The association between lymphocyte infiltration and myocardial fibrosis in cardiac transplant rejection is well known. Our gene expression data shows upregulation of complement components [26
] and CCR2 [27
] that can regulate lymphocyte function. We hypothesize that lymphocyte infiltration may be the link between the inflammatory gene profile and the myocardial fibrosis we have observed, but further experiments are required to confirm that these cells are truly lymphocytes and that this observed link is causal. Our finding are congruent with others in the field [28
], who have also shown upregulation of inflammatory/immune pathways in the aging mouse heart, although the actual genes they list are slightly different from ours. This may be because of strain differences between the studies. In a large study, Swindell [29
] combined numerous microarray studies of murine cardiac aging, and reported upregulation of numerous GO terms (over 100 biological processes were upregulated). These are also largely related to inflammation and immunity, similar to ours. The difference between our data and that of Swindell may be due to two factors: he combined data from 2 different strain of mouse, and he used 24 month-old mice, whereas we used only c57Bl6 at 18 months of age. In another study, Park and colleagues [30
] used microarray technology to assess the aging heart. They compared seven different mouse strains and found 6 GO terms were consistently upregulated across at least 6 of the 7 strains they tested. Of these 6 terms, 5 involved inflammation/immunity, and all these 5 were also upregulated in our study. Although they tested strains that were different from ours, the consistency of the finding of immune/inflammatory gene upregulation across strains strengthens our microarray findings. To our knowledge, however, ours is the first study to describe both inflammatory cell infiltration to the heart and inflammatory gene upregulation in aging mouse hearts. Future studies are required to address the links between inflammatory gene induction, cardiac fibrosis and the cardiomyopathy of aging.
The propensity of aging cardiomyocytes toward apoptosis may contribute to the known susceptibility of the elderly to diseases such as myocardial infarction and heart failure. At this early time-point in the evolution of age-related cardiomyopathy, we cannot detect differences in the number of apoptotic CM, yet we detect a significant increase in active caspase-3. The presence of active caspase 3 is considered to be a surrogate measure for apoptosis, so how can we reconcile this discrepancy? One possible explanation is that the process of apoptosis is brief, and it is relatively uncommon to see long-lived cells like cardiomyocytes undergoing apoptosis [31
]. Thus, because the event is so brief, finding true differences between groups is difficult and we are likely to underestimate the true contribution of this brief cellular event to the organ as a whole. However, our finding of similar total numbers of cardiomyocytes between young and aging mice would argue against this, and would suggest that there is truly less apoptosis. Furthermore, we believe our findings are strengthened by the use of two methods to assess apoptosis: Hairpin-1 and TUNEL. Both these methods assess for DNA cleavage associated with apoptosis, but Hairpin-1 is more specific for DNA double-strand breaks with single base 3’ overhang, and is thus thought to be more specific for apoptosis [32
]. Another reason that increased caspase-3 may not translate into increased apoptosis is that other caspase-independent pathways of apoptosis may become downregulated. To our knowledge, caspase-independent pathways to cardiomyocyte apoptosis have not previously been described in the aging murine heart. We assessed three well-characterized caspase-independent mediators of apoptosis (AIF, EndoG, and BNIP3) and found they were not suppressed compared to young hearts. Ljubicic et al recently demonstrated increased AIF in the hearts of senescent rats [33
]. There may be other caspase-independent apoptotic pathways that are downregulated to account for our findings. An alternative explanation is the upregulation of anti-apoptotic pathways that counteract the activation of caspase-3, resulting in no increase in apoptosis. We showed an increase in pAKT and Bcl-2, which may account for this, but numerous other anti-apoptotic pathways exist. Possibilities for such pathways include IAPs (inhibitor of apoptosis proteins), heat shock proteins, calpains, proteases, phosphoinositides and Bcl-2 family members, which can all antagonize caspase signaling. Other anti-apoptotic pathways downstream of caspase-3 may also be upregulated. It bears mentioning that we analyzed apoptosis and activated caspase-3 in cardiomyocytes, not in the entire tissue. Therefore our results reflect apoptosis in CM, and are not confounded by signaling in other cell types. Our finding of cardiomyocytes “primed” for apoptosis suggests that anti-apoptotic strategies are an important research direction for heart disease in the aging. This time-point, where there is a predilection towards, before there is no actual apoptosis, may be a particularly good time to intervene to prevent apoptosis.
Autophagy has been described as a part of normal aging [34
]. Whether autophagy is physiological, pathological, or a reaction to other pathologies is not known. Furthermore, whether it contributes to the cardiomyopathy of aging is also not known. We showed increased CM autophagy in the aging heart. Increased early autophagic markers beclin-1 and LC3II:I ratio (markers of the autophagosome) may represent increased autophagosome formation or decreased clearance (fusion with lysosomes i.e. autophagic flux). Here we show no impairment in the lysosome compartment by cathepsin-D staining, so the increase in autophagosomes is not due to reduction in the lysosome compartment. We did not detect differences in autophagy-related genes. Therefore, the autophagic response is likely at the posttranslational level and is more likely to be a secondary response than a constitutively activated major pathogenic cause of the cardiomyopathy. Autophagy has been associated with rescue from apoptosis in cardiomyocytes in vitro
]. It is enticing to speculate that the increases seen in anti-apoptotic and autophagic factors were able to rescue the cardiomyocytes from caspase-dependent apoptosis in vivo
Whereas studying a very old, senescent animal may show cellular and molecular changes that are secondary to heart failure, we believe that studying this time-point early in the evolution of heart failure is more likely to identify gene expression differences that are causally linked to the development of heart failure. Our findings are observational in nature, however we describe in detail the phenotype of age-related cardiomyopathy at a structural, functional, cellular and molecular level at the onset of the disease. These observations set the scene for future studies of therapies aimed at preventing or reversing the early stages of age-related heart failure.