• Log in with Facebook Log in with Twitter Log In with Google      Sign In    
  • Create Account
  LongeCity
              Advocacy & Research for Unlimited Lifespans

Photo

Patterns of Aging Biomarkers, Mortality, and Damaging Mutations Illuminate the Beginning of Aging and Causes of ...

aging lifespan mortality selection age-related cancer incidence damage mutations clock

  • Please log in to reply
No replies to this topic

#1 Engadin

  • Guest
  • 198 posts
  • 580
  • Location:Madrid
  • NO

Posted 24 January 2020 - 03:20 PM


.

 

 

 

 

F U L L   T I T L E :   Patterns of Aging Biomarkers, Mortality, and Damaging Mutations Illuminate the Beginning of Aging and Causes of Early-Life Mortality.

 

 

 

 

 

F U L L   T E X T   S O U R C E :   Cell

 

 

 

 

Highlights

 
  •  Mortality from age-related diseases is U-shaped with the nadir below reproductive age
 
  •  Quantitative biomarkers of aging change continuously throughout life
 
  •  Mutation burden causes early-life mortality and contributes to selection
 
  •  Aging is best defined by damage rather than mortality and starts very early in life
 
 
 
Summary
 
An increase in the probability of death has been a defining feature of aging, yet human perinatal mortality starts high and decreases with age. Previous evolutionary models suggested that organismal aging begins after the onset of reproduction. However, we find that mortality and incidence of diseases associated with aging follow a U-shaped curve with the minimum before puberty, whereas quantitative biomarkers of aging, including somatic mutations and DNA methylation, do not, revealing that aging starts early but is masked by early-life mortality. Moreover, our genetic analyses point to the contribution of damaging mutations to early mortality. We propose that mortality patterns are governed, in part, by negative selection against damaging mutations in early life, manifesting after the corresponding genes are first expressed. Deconvolution of mortality patterns suggests that deleterious changes rather than mortality are the defining characteristic of aging and that aging begins in very early life.
 
 
fx1_lrg.jpg
 
 
Introduction
 
Aging involves the continuous accumulation of deleterious changes, consequential loss of function, development of age-related diseases, and ultimately death. Perhaps, the most characteristic, albeit not universal (Jones et al., 2014), feature of organismal aging has been the age-related increase in frailty and mortality, i.e., “something ages if it is more likely to fall apart tomorrow than today” (Gavrilov and Gavrilova, 2004). Mortality of all common model organisms used in aging studies (mice, zebrafish, flies, nematodes, and budding yeast) and many non-model organisms (Nussey et al., 2013) follows the pattern of age-dependent growth. If mortality indeed represents aging, it would then be tempting to extrapolate the relationship between these two processes to the entire lifespan and infer the age corresponding to the initiation of the aging process by identifying the moment when age-specific mortality starts to increase.
Human aging closely follows this mortality pattern, which is thought to be a defining feature of aging. However, the early-life mortality rate in humans starts high and declines from the prenatal period to puberty, where it reaches minimum (Levitis, 2011); it subsequently starts to increase again, forming a U-shaped pattern. Some studies have reported that the nadir of mortality may lie even before reproductive age (Milne, 2006). This was widely interpreted as a support for the idea that the aging process begins around that time (Rattan, 2006); it is also thought to be consistent with evolutionary considerations, wherein the strength of natural selection remains constant during early life but declines once organisms reach the reproductive age (Kirkwood and Austad, 2000). This concept originates in the work of Hamilton (Hamilton, 1966), who thought that development and aging are parts of the same phenomenon that defines the mortality pattern and therefore that aging starts at the onset of reproduction (Levitis and Martínez, 2013).
 
The basis of the decline of mortality in early life that is widespread among humans, animals, and even unicellular eukaryotes is poorly understood (Levitis, 2011). If one defines aging as a process that leads to an increase in the probability of death with age across the whole lifetime, the process of aging would have to start at the age of minimal mortality, and organisms in their early life would develop while becoming biologically younger until the age of minimal mortality. Thus, the period of development would be accompanied by reversed aging, when levels of damage inherited or accumulated early in life is decreased with age (e.g., by repair mechanisms or damage dilution due to extensive cell proliferation and an increase in body volume during embryogenesis and early childhood). However, the relationship between aging and mortality is not set in stone. In what follows, we define aging as the accumulation of damage such as mutations, methylation changes, protein oxidation, and other deleterious changes with age (Gladyshev, 2016). Since the early research it has been thought that germline mutations drive aging (Medawar, 1946), and this concept was later applied in Hamilton’s seminal work (Hamilton, 1966) to support the conclusion that negative senescence cannot take place. However, this proposition appears to disagree with the discovery of species characterized by decreased mortality and increased fecundity over much of their lifespan (Vaupel et al., 2004).
 
Another relevant concept is the notion of trade-off between fecundity and aging. Studies have shown that decreased reproduction and the delayed onset of puberty are associated with longer lifespan (Mostafavi et al., 2017) and exceptional longevity (Tabatabaie et al., 2011), whereas early puberty has been linked to the elevated risk of heart disease and diabetes (Day et al., 2015). These findings were interpreted as supporting the antagonistic pleiotropy (Williams, 1957) and disposable soma (Kirkwood, 1977) theories of aging (Tabatabaie et al., 2011) and proving that puberty defines the start of aging. However, delayed puberty may prolong life by decelerating aging rather than by shifting its onset. Moreover, the opposite trend is observed in many species: honeybee queens live much longer than workers (Page and Peng, 2001), and in Ansell’s mole-rats, both female and male breeders live longer than non-breeders (Dammann and Burda, 2006).
 
The scientific community is currently divided over the question of when aging begins; existing opinions include conception, birth of the organism, nadir of mortality, puberty, completion of development, age of most transcriptional reversals, and perhaps other phases of human life (Allison et al., 2016). In this work, we sought to address this question quantitatively. The analysis we present points to very early life as the beginning of aging; this analysis has also led us to broader insights into the causes of early-life mortality, its role in negative selection against parental deleterious alleles, and persistence of populations in the face of high mutation rates.
 
 
Results
 
Mortality from Diseases Associated with Old Age Is U-shaped with the Nadir Below Reproductive Age
 
To evaluate the relationship between mortality and aging across the entire lifespan, we first examined overall patterns of mortality in human populations, where the most resolved data are available. Consistent with previous data (Milne, 2006), the mortality rate is U-shaped, and the nadir of all-cause mortality in developed countries lies around the age of 9 years for both sexes (Figure 1A). Men show higher mortality rates in both arms of the U-shaped curve than women. As causes of death may be different between children and adults, we were particularly interested in the mortality from diseases typically associated with aging. Interestingly, mortality patterns from heart disease, infection, and sepsis are also U-shaped, with the nadir at the age of 9 years (Figure 1B). Moreover, mortality from most other causes behaves similarly (Figures 1B and S1A), with higher mortality for men than for women. In nearly every case, the nadir is observed long before the age of puberty. Mortality resulting from causes specific to childhood shows an age-related decrease, with the maximum at birth (Figure 1C). The pattern of higher early-life mortality and the timing of its decrease are preserved when injury-related mortality is subtracted (Figure S1B).

 

 

gr1_lrg.jpg

 

 

Figure 1. U-shaped Patterns in Human Cause-Specific Mortality and Cancer Incidence

 

(A) Age-specific all-cause male, female, and total mean mortality rates.

(B) Mortality rate for major causes of death often associated with aging.
© Mortality rates for causes associated with childhood.
(D) Age-specific cancer incidence rate.
(E) Age-specific number of doctor’s office visits. Error bars represent one standard deviation from the mean.
 
 
 
As cancer incidence grows with age in adult life, we further analyzed its rate throughout the whole lifespan and found it to follow the same U-shaped pattern in both men and women (Figure 1D). The number of doctor’s office visits, even after excluding visits associated with injury and preventive care, also shows a similar pattern, with the minimum during childhood, although the nadir cannot be reliably defined (Figure 1E). Overall, we observed the U-shaped pattern across many analyzed features, and its minimum is well below the age when humans start to reproduce. The rise of mortality in the period preceding reproductive age (i.e., starting at 9 years) as well as the corresponding rise in the incidence of disease represents a challenge to the evolutionary inference of the beginning of aging after the completion of development.
 
 
Age-Related Mutations Increase throughout the Lifespan
 
To examine the nature of high early-life mortality and its relationship to the process of biological aging, we analyzed the behavior of quantitative biomarkers of aging, focusing on age-associated mutation accumulation in somatic tissues. We previously found that it can be assessed by following age-related mutations in cancers; i.e., somatic mutations in tumors that are additional relative to healthy tissues of the same patient (Podolskiy et al., 2016). In contrast to mortality rates, mutations do not show U-shaped patterns. Instead, they increase with age throughout the whole lifetime for the analyzed cancer types (Figure 2A). As the majority of age-related somatic mutations are neutral or mildly deleterious, they may be viewed as proxies for molecular damage, suggesting that damage also accumulates with age, even at pre-reproductive age. Therefore, early-life mortality appears to be unrelated to aging if the latter is defined as damage accumulation and decreased fitness. Instead, the data suggest that the decreasing mortality in early life is a disparate effect, wherein all individuals begin to grow older already in very early life and continue aging throughout the lifespan, but some die early in life from non-aging related causes, even though phenotypically, these causes (e.g., heart disease, cancer, and infectious diseases) may appear similar at young and old ages.
 
 

gr2_lrg.jpg

 

 

Figure 2. Age-Related Changes in Somatic Mutation Accumulation and DNA Methylation

 

(A) Age-specific median rate of somatic mutations for indicated cancers prevalent during childhood. Shaded areas represent 1 standard deviation obtained by the bootstrapping procedure. For malignant lymphoma, R2 = 0.56, p = 7e-6 for the whole age range, and R2 = 0.15, p = 0.4 for ages until age 18 years. For soft tissue cancer lymphoma, R2 = 0.57, p = 7e-3 for all ages, and R2 = 0.15, p = 0.34 until age 18 years. For pediatric brain cancer, R2 = 0.71, p = 4e-5 for all ages, and R2 = 2e-4, p = 0.97 until age 18 years.

 
(B) Age-dependent behavior of weighted average DNA methylation on 353 CpG sites of the human epigenetic clock. Solid black line corresponds to the best fit of weighted average methylation across samples with the same chronological age and blue shaded regions to the weighted average methylation range for samples with the same chronological age but different DNAm ages. Black dots represent individual samples. Primary data are from Horvath (2013).
 
© DNA methylation age in mice. Primary data are from Stubbs et al. (2017). The WAM range (blue shaded regions) is estimated by regression. Individual samples are represented by black points.
 
 
 
 
DNA Methylation Changes Do Not Parallel Mortality Patterns
 
One of the strongest existing biomarkers of aging is represented by the age-dependent changes in DNA methylation. Epigenetic clocks were developed for humans and mice and allow for measurement of the biological age of tissues with high precision (Horvath, 2013, Petkovich et al., 2017, Stubbs et al., 2017). The rates of epigenetic aging are associated with such parameters as sex, race, birthweight, and birth by caesarean section, as well as developmental characteristics and risk factors for aging-related diseases (Horvath et al., 2016, Marioni et al., 2019, Simpkin et al., 2016, Simpkin et al., 2017, Slieker et al., 2016). We analyzed the pace of multi-tissue epigenetic clocks in humans and mice during the period spanning the entire lifespan of these organisms. Similar to the process of cancer mutation accumulation, average DNA methylation changed monotonically in humans (Figure 2B). Consistent with Horvath (2013), the changes in weighted average DNA methylation proceeded faster during early life and slowed down during adulthood, with no evidence of U-shaped patterns. Age-dependent patterns of DNA methylation in mice appeared to mimic those in humans (Figure 2C). We further tested whether standard deviation of DNA methylation age in samples from the same age cohort for young mice and humans is larger than for adults. We found that it also grows monotonically with chronological age (Figure S2C). Taken together with the patterns of mutations, the DNA methylation data further point to continuous aging of organisms and damage accumulation in them from early development rather than from reproductive age.
 
 
Early-Life Mortality and the Role of Genotype
 
If organisms continuously and monotonically accumulate deleterious changes during early life, then there must be other mechanisms explaining the puzzling pattern of initially high and declining mortality during development. As many gene knockouts are known to lead to early-life mortality, we investigated the patterns of knockout lethality in detail. The International Mouse Phenotyping Consortium (IMPC) collected data on 2,808 individual gene knockouts in mice, 34% of which are associated with abnormal survival (Koscielny et al., 2014). Using lethality windows established for 242 knockouts from the IMPC dataset (Dickinson et al., 2016), we calculated the age-related patterns of lethality associated with gene knockouts (Figure 3A). Most lethal-knockout mice are characterized by abnormal survival before and at mid-gestation; lethality is reduced dramatically afterward, including the period after birth. As the analyzed knockout models represent particular genotypes that are viable in the parent’s heterozygous state but lethal in the homozygous offspring, the data suggest that mortality during development is explained, in part, by parental genotype (Dickinson et al., 2016, Meehan et al., 2017). Enrichment analysis of genes with lethality at different stages of development shows that mortality during early gestation is associated with regulation of transcription and translation, chromosomal organization, and regulation of the cell cycle and with regulation of organogenesis at mid-gestation (Figure S3A).
 

 

gr3_lrg.jpg

 

 

Figure 3. Causes of Early-Life Mortality

 

(A) Number of genes with lethal knockouts at different stages of development in mice.

 
(B) Heterozygous human selection coefficients for genes with knockout lethality in mice and for genes with viable knockouts in mice and humans. The vertical lines, whiskers, extend to the most extreme, non-outlier data points. These extreme data points are denoted by caps, the horizontal lines at the ends of the whiskers. The orange horizontal lines are means.
 
© Heterozygous human selection coefficients estimating selection against protein-truncating variants in the heterozygous state for genes expressed for the first time in different embryonic stages. Selection coefficients greater than 0.1 reflect loss-of-function intolerance. The vertical lines, whiskers, extend to the most extreme, non-outlier data points. These extreme data points are denoted by caps, the horizontal lines at the ends of the whiskers. The orange horizontal lines are means. The symbols above the boxes denote significance: ∗∗∗∗: p < 0.0001, ∗∗∗: p < 0.001, ∗∗: p < 0.01, ∗∗∗ p < 0.05, -: p-value is not significant.
 
(D) Age-specific perinatal mortality rate.
 
(E) Spontaneous abortion as a function of maternal age in humans (left), and fraction of spontaneous abortions characterized by abnormal karyotype as a function of maternal age (right).
 
 
Similar to the evolutionary costs of complete gene knockout, there is an active selection against heterozygous loss of gene function, which can be assessed by the distribution of selection coefficients for heterozygous protein-truncating variants (Cassa et al., 2017). Moreover, lower heterozygous selection coefficients are associated with decreased severity of associated diseases and later age of their onset. The frequencies of deleterious alleles change according to the theory of evolution of dominance (Fisher, 1931, Wright, 1931). When the effect of the heterozygous loss of function on fitness is stronger than the rate of genetic drift, these alleles are selected against in the population. This makes homozygotes for this allele occur extremely rarely, and selection happens almost entirely through heterozygotes. We found that selection against heterozygous knockouts of human genes orthologous to lethal mouse homozygous knockouts is such that it remains high in the prenatal stage and decreases afterward, being lowest for the knockouts that are viable in mice or humans (Figure 3B). This observation and the well-established contribution of consanguinity to early-life mortality in human populations further point to the role of parental genotype in selection against deleterious alleles during development.
 
The decreasing lethality during development is consistent with the decrease in the number of essential genes that are expressed for the first time during development (Figure S3B). Thus, it appears that the damaging alleles start causing deleterious changes that lead to lethality when they are first expressed. Indeed, we calculated selection coefficients for genes expressed for the first time at each stage of development and found that the coefficients are higher during the early stages and decrease afterward (Figure 3C). When genes expressed during embryonic development were clustered, the main clusters corresponded to the genes first expressed very early during embryogenesis and at mid-gestation (Figure S3C).
 
 
 
 
 
.../...
 
 
 
 
 
 
 
 
 
 
.

Edited by Engadin, 24 January 2020 - 03:21 PM.






Also tagged with one or more of these keywords: aging, lifespan, mortality, selection, age-related, cancer, incidence, damage, mutations, clock

0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users