Elsevier

Biosystems

Volume 104, Issues 2–3, May–June 2011, Pages 94-98
Biosystems

Compositional perspectives on human brain aging

https://doi.org/10.1016/j.biosystems.2011.01.005Get rights and content

Abstract

Using publicly available microarray data from the frontal cortex of 30 individuals, spanning the ages of 26–106 years old, we investigate the expression patterns of compositionally distinct genes during human brain aging. Our analyses revealed that at advance ages, GC-poor genes appear to be induced while GC-rich genes are repressed. Interestingly, investigations upon two different types of genes, named pivotal (permanently expressed genes) and non-pivotal (on-off regulated genes), revealed an induction of the GC-poor pivotal genes and a repression of the GC-richer non-pivotal genes at advanced ages. Summarizing, this study shows that genes with different compositional properties have opposite age-related expression patterns, suggesting an implication of different regulation mechanisms related to their localization in different chromatin structure, which correlates with the GC level. Finally, an innovative approach on investigating human aging process is suggested, which involves the base composition of genes.

Introduction

In contrast to the genomes of cold-blooded vertebrates, the genomes of warm-blooded vertebrates are characterized by a strong compositional heterogeneity. The human genome, has a compositional pattern that is a mosaic of isochores, long DNA stretches (average size 0.9 Mb) having a fairly homogeneous composition. Isochores belong to 5 families that cover a broad GC range (30–60%), named L1, L2, H1, H2 and H3 (in increasing order of GC level; Bernardi, 2005, Bernardi, 2007, Costantini et al., 2006). The two GC-richest families, H2 and H3, which represent approximately 15% of the genome, comprise approximately 50% of the protein-coding genes. The high gene density of GC-rich isochores is accompanied by other striking properties such as open chromatin structure, location at the periphery of the interphase nucleus, high density of SINES (short interspersed nuclear elements), low density of LINES (long interspersed nuclear elements), an early replication, a high recombination level, a high mutation rate and more CpG un-methylated targets (Bernardi, 2005, Varriale and Bernardi, 2010). The properties of the GC richest family are opposite to those found in the remaining 85% of the genome (low gene density) represented by the GC-poor isochores families.

Increasing evidence suggests that GC-poor isochores are implicated during development. Notably, genes expressed during the early developmental stages of the mouse hold AT-ending optimal codons and have lower GC levels compared to the genes expressed in the late stages (Ren et al., 2007). Moreover, development-pivotal (up/down-regulated) and development-specific (on-off regulated) genes were found to contain specific GC patterns, suggesting the action of different developmental regulatory mechanisms due to their chromatin structure, which correlates with the GC level. Similarly, before and after the differentiation of mouse embryonic stem cells to neural precursor cells, most expression changes occur on genes localized in LINE-rich and GC-poor regions (Hiratani et al., 2004).

With this background, we took advantage of publicly available expression data and approached the fairly complex of brain aging in relation to the base composition of the involved genes. More specifically, our investigation is focused on a relation between aging process and GC level of genes, through the estimation of the age-related expression patterns for three distinct compositional classes of genes.

Section snippets

Material and methods

The microarray data (as inferred from Affymetrix arrays) from a study that investigated frontal cortex aging using samples from 30 individuals, which spanned the ages of 26–106 years old (Lu et al., 2004, GEO series GSE1572) was retrieved from the GEO database (Gene Expression Omnibus, http://www.ncbi.nlm.nih.gov/geo/). Using the GenBank accession numbers assigned to the probes of the U95Av2 array, nearly 6000 complete coding sequences were extracted as in a previous work (Arhondakis et al.,

Average expression patterns of each compositional class during frontal cortex aging

First, the experiments (30 experiments from 30 individuals representing 28 different ages) were arranged in increasing order of age. For each age the average expression level for each compositional class (GC-poor, GC3 <45%; GC-medium, GC3 from 45 to 65%; GC-rich, GC3 ≥65%) was estimated, taking into account the expressed genes in each sample. Pairwise comparisons of the average expression for the same compositional classes between the nearest neighbors’ individuals (in terms of age; Fig. 1A,

Discussion

In this work, human brain aging has been approached in relation to the compositional properties of human genes. The analyses for the three compositional classes were conducted at four levels: (i) the average expression between adjacent in age individuals (nearest neighbors in term of age), (ii) the average expression between groups of ages, (iii) at the single gene level (age-related patterns) and (iv) fold-changes in genes expression between old and young individuals. The first analysis

Acknowledgments

We would like to express our gratitude to Professor Giorgio Bernardi for all his valuable pieces of advice and support throughout this work. We also thank Oliver Clay for constructive criticisms and Ethan Ford for critical reading of the manuscript. In addition we acknowledge institutional funds.

References (18)

There are more references available in the full text version of this article.

Cited by (5)

  • Principles of brain aging: Status and challenges of modeling human molecular changes in mice

    2021, Ageing Research Reviews
    Citation Excerpt :

    Importantly, age-related neural dedifferentiation has been extensively observed in humans, in several cortical regions, and it is believed to negatively correlate with cognitive performance (Koen and Rugg, 2019). The mechanisms of dedifferentiation are not known, although one speculation is that in the aged human brain there is an epigenetic shift that tends to increase the expression of genes rich in AU-ending codons (Arhondakis and Kossida, 2011), which indeed correlate with a less differentiated phenotype (Fornasiero and Rizzoli, 2019). When young mice (2 months old) are compared to aged mice (22 months old) the expression level of the synaptic vesicle protein synaptotagmin I seems decreased in aged mice (Jiang et al., 2001).

  • Aging dependent effect of nuclear tau

    2017, Brain Research
    Citation Excerpt :

    It is also known the dramatic nuclear change takes place between aging neurons and AD neurons, possibly related to an aberrant attempt to re-enter into the cell cycle associated with aging stress (Zhu et al., 1999) and these events may begin the neurodegenerative process associated with age (McShea et al., 2007). In this sense, the changes observed in the nuclear tau distribution in aging cells with respect to younger cells, and even more so with AD neurons, could be associated to modifications of the chromatin organization determining alterations in gene expression (Tan et al., 2010; Arhondakis and Kossida, 2011). A number of studies have demonstrated the presence of non-phosphorylated tau (Tau-1) in the nucleolus (Loomis et al., 1990; Thurston et al., 1996, 1997; Sjoberg et al., 2006; Rossi et al., 2008), and now we observed for the first time a nucleolar location of AT100 in the pyramidal cells of the CA1 region, that could indicate a further involvement in chromatin organization, in this case related to the rDNA sequences, as the loss of nucleolar heterochromatin has been shown to trigger accelerated aging and that heterochromatin prevents recombination between rDNA repeats, thereby preserving the nucleolar structure (Sinclair and Guarente, 1997; Peng and Karpen, 2007).

  • Transcriptome map of mouse isochores in embryonic and neonatal cortex

    2013, Genomics
    Citation Excerpt :

    This finding, although not possible to ascertain due to the low number of developmental stages, is in line with the observation of Ren et al. [20], who reported that the genes expressed in the early developmental stages, compared to the genes expressed in the later developmental stages, have more AT-ending codons (lower GC). The functional role of genes' compositional properties on gene expression activity, in different cellular contexts (see Introduction section) and biological processes, i.e., aging [26], and that of the isochores' on the regulation mechanisms, represents a relevant subject of study. Indeed, a recent work [27], reported differences in the frequency of the short sequences (di- and tri-nucleotides) between the isochore families, accounting for their different properties, i.e., codon frequencies, the distribution of DNA methylation, CpG doublets, and CpG islands [28], as well as the chromatin structure [29–31].

View full text