Younger age of cancer initiation is associated with shorter telomere length in Li-Fraumeni syndrome

Li-Fraumeni syndrome (LFS) is a cancer predisposition syndrome frequently associated with germ line TP53 mutations. Unpredictable and disparate age of cancer onset is a major challenge in the management of LFS. Genetic modifiers, including the MDM2-SNP309 polymorphism, and genetic anticipation have been suggested as plausible explanations for young age of tumor onset, but the molecular mechanisms for these observations are unknown. We speculated that telomere attrition will increase genomic instability and cause earlier tumor onset in successive generations. We analyzed mean telomere length and MDM2-SNP309 polymorphism status in individuals from multiple LFS families and controls. A total of 45 peripheral blood lymphocyte samples were analyzed from 9 LFS families and 15 controls. High rate of MDM2-SNP309 was found in TP53 carriers (P = 0.0003). In children, telomere length was shorter in carriers affected with cancer than in nonaffected carriers and wild-type controls (P < 0.0001). The same pattern was seen in adults (P = 0.002). Within each family, telomere length was shorter in children with cancer than in their nonaffected siblings and their noncarrier parents. Telomere attrition between children and adults was faster in carriers than in controls. Our results support the role of MDM2-SNP309 as a genetic modifier in LFS. The novel finding of accelerated telomere attrition in LFS suggests that telomere length could explain earlier age of onset in successive generations of the same family with identical TP53/MDM2-SNP309 genotypes. Furthermore, telomere shortening could predict genetic anticipation observed in LFS and may serve as the first rational biological marker for clinical monitoring of these patients.     

Reconstructing ancestral gene content by coevolution

Inferring the gene content of ancestral genomes is a fundamental challenge in molecular evolution. Due to the statistical nature of this problem, ancestral genomes inferred by the maximum likelihood (ML) or the maximum-parsimony (MP) methods are prone to considerable error rates. In general, these errors are difficult to abolish by using longer genomic sequences or by analyzing more taxa. This study describes a new approach for improving ancestral genome reconstruction, the ancestral coevolver (ACE), which utilizes coevolutionary information to improve the accuracy of such reconstructions over previous approaches. The principal idea is to reduce the potentially large solution space by choosing a single optimal (or near optimal) solution that is in accord with the coevolutionary relationships between protein families. Simulation experiments, both on artificial and real biological data, show that ACE yields a marked decrease in error rate compared with ML or MP. Applied to a large data set (95 organisms, 4873 protein families, and 10,000 coevolutionary relationships), some of the ancestral genomes reconstructed by ACE were remarkably different in their gene content from those reconstructed by ML or MP alone (more than 10% in some nodes). These reconstructions, while having almost similar likelihood/parsimony scores as those obtained with ML/MP, had markedly higher concordance with the coevolutionary information. Specifically, when ACE was implemented to improve the results of ML, it added a large number of proteins to those encoded by LUCA (last universal common ancestor), most of them ribosomal proteins and components of the F₀F₁-type ATP synthase/ATPases, complexes that are vital in most living organisms. Our analysis suggests that LUCA appears to have been bacterial-like and had a genome size similar to the genome sizes of many extant organisms.The problem of reconstructing ancestral states is as old as the field of molecular evolution, pioneered by Fitch around 40 yr ago (Fitch 1971). This first algorithm assumed a binary alphabet and was based on the maximum parsimony (MP) criterion, i.e., find the labels to the internal nodes of a tree that minimize the number of changes or mutations along the tree edges. Over the years this basic algorithm was generalized in many ways. Sankoff (1975) showed how to efficiently solve versions of the MP problem with a nonbinary alphabet and with multiple edge weights. Algorithms for inferring ancestral sequences based on the maximum likelihood (ML) principle (instead of MP) were suggested more than 15 yr later (Barry and Hartigan 1987; Felsenstein 1993; Pagel 1999; Pupko et al. 2000; Krishnan et al. 2004; Elias and Tuller 2007), aiming to identify the ancestral sequences that are most likely, given the current data in probabilistic terms. In a manner analogous to inferring an ancestral sequence such as a protein or gene (Pupko et al. 2000; Blanchette et al. 2004; Ma et al. 2006), many studies have aimed to reconstruct ancestral genomes (genomic content or gene content; e.g., see Boussau et al. 2004; Ouzounis et al. 2006; Putnam et al. 2007). Most existing phylogenetic reconstruction algorithms can be adapted to reconstructing ancestral gene content (e.g., see Felsenstein 1993; Yang 1997; Swofford 2002).The main problem related to reconstructing ancestral sequences and gene inventories is that, in practice, the reconstructed sequences often contain a large number of errors. A major source of this phenomenon is the existence of multiple local and/or global maxima (i.e., solutions that are different but have the same scores of “goodness”) in the solution space searched by both the ML and the MP approaches (e.g., see Fig. 1; Chor et al. 2000). Furthermore, due to the statistical nature of the problem, and as both ML/MP assume that different sites and different genes/proteins evolve independently, increasing the amount of information used (the lengths of the sequences and the number of organisms) (Li et al. 2008) does not guarantee a decrease in the error rate. Thus, in many cases, the confidence that we may assign to the most likely or most parsimonious reconstructed ancestral state is not very high.Open in a separate windowFigure 1.A simple example demonstrating how coevolution can be used to improve ancestral sequence reconstruction. This example includes a coevolving forest with two trees, the tree edges are solid lines, while the coevolutionary edges are dashed wavy arrows; the ancestral states (e.g., the labels at the internal nodes x1 and y1) are smaller, while the known nonancestral states are larger and in italics (for formal definitions, see Supplementary Note 1). (A) The reconstructed ancestral states of two proteins (protein x and protein y); for protein x there are two parsimonious solutions: In one solution, all of the labels of the three internal nodes, t1, t2, t3, are “0”; in the second solution, all of the labels of the three internal nodes are “1.” (B) By taking into consideration the coevolution of protein x with protein y, the “all ‘1’ ” solution is chosen for x, thus resolving the ambiguity in the labels of x by using information about protein y with which it interacts, and which has less ambiguous labels.In this work we describe a novel approach for improving the accuracy of reconstructed ancestral genomes. Our approach is based on utilizing information embedded in the coevolution of interacting proteins. The potential utility of this strategy is intuitively simple: If the ancestral reconstruction of protein A is ambiguous (e.g., different solutions achieve similar scores), but the ancestral reconstruction of protein B is unambiguous, then, knowing that A and B have coevolved (usually due to some functional interaction) can help us disambiguate the ancestral reconstruction of A. A simple illustration of this idea is described in Figure 1. As demonstrated further on, our approach can successfully distinguish between solutions with similar parsimony/likelihood scores.Importantly, coevolutionary relations between proteins are quite ubiquitous and have been traced and reported in numerous studies (Pazos et al. 1997; Chen and Dokholyan 2006; Marino-Ramirez et al. 2006; Barker et al. 2007; Wapinski et al. 2007; Felder and Tuller 2008; Tuller et al. 2009). One such source is protein–protein interactions (PPIs): Several studies have already demonstrated a link between coevolution and physical interactions. These investigations used the fact that interacting proteins tend to coevolve in order to successfully predict physical interactions (e.g., see Wu et al. 2003; Sato et al. 2005; Juan et al. 2008). Here, we take the opposite approach and use data on physical interactions between proteins as a proxy for their coevolution. Large-scale PPI datasets currently exist only for relatively few organisms, but as physical interactions tend to be conserved across species, at least to some extent, (e.g., see Wu et al. 2003; Liang et al. 2006; Hirsh and Sharan 2007), it is plausible to assume that existing PPIs testify to coevolutionary relations across the ancestral tree. Another source of coevolution information that we utilize is metabolic adjacency: Following the studies of Spirin et al. (2006) and Zhao et al. (2007) we assume that metabolic enzymes that catalyze consecutive reactions in the same pathway tend to coevolve. We also incorporate additional sources of coevolution information, including genomic proximity, coexpression, and gene fusion, which have all been shown to be good predictors of coevolution (Lee et al. 2004; Chen and Dokholyan 2006; Jensen et al. 2009; Tuller et al. 2009). The coevolution data from all of these sources is combined to reconstruct the evolutionary history of 95 unicellular bacteria, archaea, and eukaryotes.     

Neural correlates of delayed visual–motor performance in children treated for brain tumours

Both structural and functional neural integrity is critical for healthy cognitive function and performance. Across studies, it is evident that children who are affected by neurological insult commonly demonstrate impaired cognitive abilities. Children treated with cranial radiation for brain tumours suffer substantial structural damage and exhibit a particularly high correlation between the degree of neural injury and cognitive deficits. However the pathophysiology underlying impaired cognitive performance in this population, and many other paediatric populations affected by neurological injury or disease, is unknown. We wished to investigate the characteristics of neuronal function during visual–motor task performance in a group of children who were treated with cranial radiation for brain tumours. We used Magnetoencephalography to investigate neural function during visual–motor reaction time (RT) task performance in 15 children treated with cranial radiation for Posterior Fossa malignant brain tumours and 17 healthy controls. We found that, relative to controls, the patient group showed: 1) delayed latencies for neural activation in both visual and motor cortices; 2) muted motor responses in the alpha (8–12 Hz) and beta (13–29 Hz) bandwidths, and 3) potentiated visual and motor responses in the gamma (30–100 Hz) bandwidth.Collectively these observations indicate impaired neural processing during visual–motor RT performance in this population and that delays in the speed of visual and motor neuronal processing both contribute to the delays in the behavioural response. As increases in gamma activity are often observed with increases in attention and effort, increased gamma activities in the patient group may reflect compensatory neural activity during task performance. This is the first study to investigate neural function in real-time during cognitive performance in paediatric brain tumour patients.