-Omics and aging: from biomarkers to systems biology (2)

Jonas Zierer et al. Integration of ‘omics’ data in aging research: from biomarkers to systems biology 

Translated by Evgenia Ryabtseva

(Continuation, the beginning of the article is here.)

Valdes et al. (2013) thoroughly analyzed the application of new technologies for the evaluation of various "-omics" for the identification of molecular markers of aging. Therefore, the following section provides only a brief description of the key results, and focuses on the most recent data.

Modern chips allow analyzing up to 5 million single nucleotide polymorphisms or "snips" (from the English single nucleotide polymorphisms, SNPs) (SNPs) (Ha et al., 2014). Today, the latest generation sequencing technology is slowly replacing chip technology, which has reduced sequencing costs to less than $0.10 per million nucleotide base pairs (Liu et al., 2012). Therefore, to date, genetic variations can be evaluated with a resolution of up to a single nucleotide.

While longevity itself is considered to be only 20% heritable (Murabito et al., 2012), for many age-related diseases this indicator is much higher. For example, the heritability of Alzheimer's disease is above 70% (Gatz et al., 2006), and osteoarthritis and cataracts – 50% (Hammond et al., 2001).

The GenAge database contains approximately 300 human genes potentially involved in aging, the selection of which is based on homology with the genes of model organisms (Tacutu et al., 2013). Sebastiani et al. (2012) recently published an improved model of 281 snips, which made it possible to distinguish centenarians from younger people of the control group in a cohort of 1,715 people. One of these snips is located in the apolipoprotein E gene, which is currently the only gene reliably associated with longevity at the genome-wide significance level (Deelen et al., 2011; Nebel et al., 2011). Common genetic variants at this locus are associated with accelerated aging and loss of cognitive function (Johnson, 2006; Davies et al., 2014), possibly due to an increased risk of coronary heart disease, stroke and Alzheimer's disease (Smith, 2002). Even despite the evidence provided by a number of studies that mutations of transcription factors of the FOXO family are also associated with longevity (Willcox et al., 2008; Flachsbart et al., 2009), when conducting genome-wide association studies, these results could not be reproduced at the genome-wide significance level.

The most common epigenetic mechanism is DNA methylation, which is considered one of the main reasons for the suppression of gene expression. Unlike the genome, which is the same for all cells of the body, the epigenome is an important factor in cell differentiation, leading to the appearance of pronounced epigenetic differences between cells of different types (Meissner, 2010). A state-of-the-art methylation assessment chip from Illumina allows analyzing more than 485,000 methylation zones and covers 99% of the RefSeq database genes (Illumnia, 2011). However, it covers less than 10% of variable regions (Ziller et al., 2013).

Epigenome is influenced by environmental and lifestyle factors (Nakajima et al., 2010; Alegría-Torres et al., 2011; Breitling et al., 2011), and is also associated with many complex diseases, such as neurodegenerative diseases (Portela & Esteller review, 2010) and cancer (Ehrlich, 2002; Horvath, 2013). Nearly 500 differentially methylated regions of the genome associated with chronological age and age-related changes in phenotype characteristics, such as respiratory function, blood cholesterol levels and maternal longevity, have been identified (Bell et al., 2012). A recent study by Weidner et al. (2014) showed that the methylation profiles of three zones are sufficient to determine the chronological age. Therefore, many of the previously identified methylation zones can be independently associated with age. An interesting fact is that the change in the level of methylation depending on age is uniform for many tissues and cell types (Horvath, 2013). Together, these patterns form a global model of hypomethylation of repetitive sequences, hypomethylation of promoter regions and increased intracellular variability (Cevenini et al., 2008; Bacalini et al., 2014). In addition to DNA methylation, other epigenetic changes are associated with the longevity of model organisms, such as histone methylation and acetylation (Dang et al., 2009; Greer et al., 2010). The study of these modifications in the human genome may shed light on the currently unknown mechanisms of aging.

The whole complex of RNA transcripts is called a transcriptome. It can be divided into coding RNAs that subsequently enter into the process of protein translation, and non-coding RNAs that perform various functions, such as regulating gene expression (Eddy, 2001). The number of transcripts can be measured using chips or sequencing methods.

Just like the epigenome, gene expression changes significantly with age. In a pioneering study comparing frontal cortex tissue samples from 30 people of different ages, 463 differentially expressed genes were identified (Lu et al., 2004). Despite the small sample size, the results obtained were reproduced in subsequent experiments. Four years later, Berchtold et al. (2008) identified several thousand age-related changes in gene expression in four different brain tissues. Later studies by different research groups revealed pronounced age-related transcriptome changes in other tissues, such as skin, adipose tissue (N = 865) (Glass et al., 2013) and kidneys (N = 134) (Rodwell et al., 2004). Most of these changes in different tissues do not overlap. Meta-analysis of data on different species and tissues revealed only 73 genes stably associated with age (de Magalhães et al., 2009). This indicates that most of the detected age-related changes in the transcriptome are species-specific, or are false-positive data (review by Valdes et al., 2013). Meta-analysis revealed that genes associated with immune response and lysosomes tend to be overexpressed in the elderly, whereas genes associated with mitochondria and oxidative phosphorylation tend to be underexpressed (de Magalhães et al., 2009).

Due to alternative splicing and post-translational modification of protein molecules, the estimated number of proteins exceeds the number of genes by two orders of magnitude (Ginsburg & Haga, 2006). However, modern proteomics technologies based on immunological analysis, protein microarrays or mass spectrometry allow measuring only a small fraction of the proteome (up to 1,000 proteins in a sample). The most complete description of the human proteome for various tissues today consists of 18,097 proteins (19,376 isoforms) identified during ten thousand experiments using mass spectrometry (Wilhelm et al., 2014).

Because of these technical problems, proteomic research in the field of aging is still devoted to the study of small protein complexes in small-sized samples. In an early paper on the protein composition of the lateral broad thigh muscle, Gelfi et al. (2006) revealed the predominance of several proteins involved in aerobic metabolism and a low content of anaerobic metabolism proteins in tissue samples of elderly people. In addition, the elderly were characterized by a consistently low expression of six transport proteins. However, in this study, only 12 tissue samples were analyzed without repetitions. In a recent study conducted by the review authors, SOMAscan technology was used to analyze more than 1,000 proteins in 200 blood plasma samples (Menni et al., 2015). As a result, a pronounced association of eleven proteins with chronological age, as well as age-related changes in the phenotype, such as respiratory function and blood pressure, was revealed. These results were replicated using an independent cohort.

Even in the absence of detailed proteomic studies, there is a high probability of the existence of a relationship between proteins and a number of age-related diseases. For example, cardiovascular diseases (Mehra et al., 2005) and Alzheimer's disease (Swardfager et al., 2010) are consistently associated with elevated levels of proinflammatory cytokines.

There are many different modifications, ranging from the attachment of small (e.g., acetylation or phosphorylation) and large molecules, such as lipid or carbohydrate chains (e.g., palmitoiling, glycosylation), to the attachment of whole proteins (e.g., ubiquitination).

The most common modification is glycosylation, which consists in attaching a carbohydrate chain to a protein molecule. Attached oligosaccharides – glycans – presumably act as structural elements of proteins or specific binding zones for other glycans or proteins (Varki et al., 2009). However, glycans are very diverse, and many of them have not yet been characterized and annotated. Therefore, glycans can have many additional functions. For example, in the intestine, glycans act as food for microorganisms (Koropatkin et al., 2012), which may be involved in immune function, which plays an important role in aging processes. A recently developed high-performance technique allows simultaneous measurement of a large number of glycans both in one protein and in all proteins simultaneously (Royle et al., 2008; Pucić et al., 2011).

The application of this technology to epidemiological cohorts has shown that the structures of glycans are stable over time for each individual (Gornik et al., 2009), but are very diverse within the population (Knezević et al., 2009; Pucić et al., 2011). Differences in glycomes characteristic of different types of cancer were also found (Fuster & Esko, 2005; Adamczyk et al., 2012). Kristic et al. (2013) demonstrated the existence of a strong association between immunoglobulin G glycans and age – a linear combination of three glycans explained 58% of the observed variability associated with chronological age in a study on four independent populations with a total of 5,117 participants.

The studied molecules are often called metabolites, since many of them are intermediate or final products of cellular metabolism. To date, the human metabolome database (Wishart et al., 2013) contains more than 40,000 different metabolites from different tissues. Just like proteomics, at the present stage, metabolomics does not have at its disposal an analytical method that allows identifying all metabolites and quantifying them in a single experiment. Modern platforms based on either chromatography in combination with mass spectrometry or nuclear magnetic resonance allow for the approximate measurement of thousands of metabolites in the analysis of unknown quantities and somewhat less when using specified directional approaches. The limitations of the directional approach, on the other hand, provide the advantage of higher sensitivity, as well as the possibility of absolute (and not only relative) quantitative determination of the compound, as well as its direct identification (Patti et al., 2012; Tzoulaki et al., 2014).

In 2008, as part of the first full-scale study of associations with age, a plasma metabolome of 269 people was analyzed using a non-directional approach. The authors demonstrated the existence of a correlation with chronological age for 100 out of 300 compounds (Lawton et al., 2008). Somewhat later, larger cohorts were used to study associations between metabolite levels and age using predefined and undefined metabolomic platforms. Yu et al. (2012) analyzed 131 predefined metabolites in 2,162 KORA study participants, while the authors of this review analyzed 280 unidentified metabolites in 6,055 twins from the TwinsUK study cohort (Menni et al., 2013b). In both studies, it was found that the levels of half of the analyzed metabolites were associated with chronological age. It also turned out that for many of these metabolites there is a pronounced correlation with age-related changes in phenotype characteristics, such as respiratory function, bone mineral density and cholesterol concentration in the blood (Menni et al., 2013b), Alzheimer's disease (N = 93) (Orešič et al., 2011), cancer (reviewed by Teicher et al., 2012) and type 2 diabetes mellitus (N = 100) (Suhre et al., 2010; Menni et al., 2013a). One of these metabolites, C–glycosyltryptophan, is a potential degradation product of glycosylated proteins.

The largest microbial community inhabits the digestive tract, in which the number of microbial cells and their genes exceeds the number of human cells and their genes by 10 and 100 times, respectively (Peterson et al., 2009; Zhu et al., 2010; The Human Microbiome Project, 2014a). The Human Microbiome Project has identified more than 10,000 different species of microorganisms with millions of protein-coding genes (Turnbaugh et al., 2007; Peterson et al., 2009; Biagi et al., 2012), with the genomes of more than 1,000 of these microorganisms to date were completely sequenced (The Human Microbiome Project, 2014b). Despite the fact that studies involving twins have demonstrated a moderate genetic influence on a number of taxometric types of microorganisms, most of the variations are due to environmental influences (Goodrich et al., 2014).

The composition of microflora varies significantly for different people (Turnbaugh et al., 2007; Zhu et al., 2010) and even for different parts of the body (Kong, 2011). It has a huge impact on many biological processes, such as immune responses, metabolism and the course of diseases (Zhu et al., 2010; Grice & Segre, 2012). Whereas the microbiome is characterized by relative stability in adulthood, it changes significantly at later stages of life (Guigoz et al., 2008; Biagi et al., 2010; Claesson et al., 2011). Biagi et al. (2010) observed pronounced changes in the gut microbiome of centenarians compared to young adults, as well as with older people, which is manifested by a loss of diversity and an increase in the number of rods and protobacteria. There is evidence that, under certain conditions, the latter contribute to the development of inflammation (Round & Mazmanian, 2009). Similar observations were made in other populations of elderly people, and were recognized as a manifestation of the dietary and living conditions of elderly patients (Claesson et al., 2012).

During epidemiological studies, thousands of clinically significant characteristics of phenotypes that do not fall under the criteria of data types described by "-omics" were collected. They range from anthropometric indicators to health and lifestyle questionnaires (Moayyeri et al., 2013). The collection of multilateral clinical data is extremely important for the identification of gene pleiotropy and interactions between clinical phenotype changes, such as combined diseases (Houle et al., 2010). The development of technologies "-omik" puts at the disposal of scientists statistical and bioinformatic methods of analysis of multilateral data. This makes it possible to simultaneously study numerous clinical changes in phenotypes, thus defining a new direction of phenomics (Houle et al., 2010).

Phenomics is of exceptional importance for the study of aging. Dozens of clinical changes in phenotypes, such as Parkinson's disease (Reeve et al., 2014), Alzheimer's disease (McAuley et al., 2009), body mass index, blood pressure (Mungreiphy et al., 2011) and bone mineral density (Warming et al., 2002), as well as lifestyle parameters such as nutrition quality (Wieser et al., 2011), smoking and physical activity demonstrate a pronounced relationship with age (Harman, 1988; Wang et al., 2009). Complex indicators, such as the Rockwood senility Index (Rockwood & Mitnitski, 2007), combine several similar clinical parameters to form a more homogeneous complex of changes in the phenotype – senility – from its various manifestations. Such indicators of senility can be considered as parameters of biological age (Mitnitski et al., 2013). Many of these (and other) clinical characteristics of phenotypes correlate or even depend on each other (McAuley et al., 2009; Baylis et al., 2014). Only an extensive collection of data and their comprehensive analysis will help to uncover these relationships and identify cause-and-effect relationships.

Continuation: From "-omik" to systems biologyPortal "Eternal youth" http://vechnayamolodost.ru

30.10.2015