Monthly Archives: October 2012

Beating a Toxin-Antitoxin System; Evading Suicide

Bacteria have evolved many different systems to evade viral predation. One strategy, abortive infection (Abi), involves altruistic suicide. Mediated by a toxin-antitoxin (TA) system, the suicide of the infected cell protects the clonal bacterial population by preventing the spread of replicated bacteriophage. A new paper in Plos Genetics has discovered a molecular mimicry-based strategy that allows phage to escape abortive infection.

Toxin-antitoxin systems are widespread prokaryotic genetic elements found on both plasmids and bacterial chromosomes. Encoding a relatively long-lived toxin and a more labile antitoxin expressed from a single bi-cistronic operon, they were originally characterised as ‘addiction modules’. In the event that a plasmid expressing a TA system fails to be inherited by a daughter cell, the absence of antitoxin allows the persisting toxin to kill the cell – post-segregational killing. These attributes as ‘selfish elements’ made it slightly surprising that so many TA systems have been found encoded on bacterial chromosomes themselves. I’ve previously written about an example of one such TA system’s activity in mediating a stress response in E. coli, and they’ve also been implicated in the formation of antibiotic resisting ‘persister’ cells.

TA systems have been classified into three different classes defined by the level of the molecular interaction between their two components. In type I systems, translation of the toxin is prevented by an antisense RNA antitoxin binding to its’ transcript, whilst in type II systems both partners are proteins. Most recently, type III TA systems have been characterised in which the toxin is neutralised by binding of an RNA antitoxin. Examples of all three varieties have been found to protect bacteria from phage infection via abortive infection; phage replication disrupts the normal cellular transcriptional program, interrupting antitoxin production and hence leading to cell death.

The first type III TA system to be characterised, ToxIN, was found on plasmids in the phytopathogen, Pectobacterium atrosepticum (Pba), and shown to inhibit the propagation of multiple different bacteriophage. ToxN is an endoribonuclease, whilst the antitoxin ToxI, is a 36nt RNA structured as a ‘pseudoknot’. The partners combine into a hetero-hexameric structure composed of 3 ToxN molecules and 3 ToxI pseudoknots.

Blower et al. have discovered that phage can evade the Abi system by producing molecular mimics of the ToxI RNA. The lytic bacteriophage ΦTE, normally fails to infect Pba carrying a toxIN-containing plasmid. At low frequency however, new phage strains emerge capable of evading the Abi system. Upon sequencing the genomes of these ‘escape strains’, the researchers discovered that they all contained sequence expansions at one specific locus. The toxI locus contains 5.5 repeats of the 36nt RNA pseudoknot-encoding sequence. The ‘escape locus’ from the phage normally encoded 1.5 repeats of a pseudo-ToxI sequence. In all the escape strains this repeat had been expanded so that it contained either 4.5 or 5.5 repeats. These expansions had probably arisen due to strand-slippage during phage replication. In one escape strain homologous recombination had occurred between the phage pseudo-ToxI and the endogenous toxI; the phage had effectively hijacked a normal antitoxin- encoding gene.

The 1.5 repeat pseudo-ToxI could not inhibit Abi (as the sequence was out of phase it did not actually encode a functional psudoknot). However, the repeat expansions had allowed the phage to make an antitoxin mimic that protected them from the TA system and hence Abi.

ΦTE is capable of generalised transduction – the ability to package and transfer chromosomal and plasmid DNA from its’ host and transfer it during infection. Blower et al. showed that one of the ΦTE escape strains is able to transduce the plasmid encoded ToxIN – a case of a bacteriophage horizontally transferring an anti-phage defence mechanism. This brings into focus the complex evolutionary dynamics operating between the three different genetic entities being studied; the bacterial cell, the plasmid encoding the TA system, and the bacteriophage evading it and potentially propagating it. From the selfish viewpoint of the TA module what’s best, preventing the spread of the phage or being disseminated by it? These speculations aren’t about to be easily answered, however, it is an interesting way to analyse further examples of similar systems.

Blower, T., Evans, T., Przybilski, R., Fineran, P., & Salmond, G. (2012). Viral Evasion of a Bacterial Suicide System by RNA–Based Molecular Mimicry Enables Infectious Altruism PLoS Genetics, 8 (10) DOI: 10.1371/journal.pgen.1003023

Barcoding the Brain

A new DNA-sequencing based idea for mapping the connectome is presented in Plos biology.

The defining purpose of neurons is information transmission and processing within a network. Hence, to appreciate neural function we must look at the interactions between neurons; understand their connections; know which other neurons they synapse with.

The idea of documenting all the connections within brains – mapping the ‘connectome’ – is receiving a lot of attention (and money) at the moment. However, the only comprehensive technique so far available is the painful reconstruction of the synaptic map from electron micrographs of serially sectioned brains. Seeing as the human brain is estimated to contain 85 billion neurons making 1014/1015 synaptic connections, one does have to wonder whether even if we could accurately document the connectome and afford it, would we really appreciate it?

The only organism to have had it’s connectome documented in this way is the nematode worm, C. elegans. Mapping its’ 302 neurons and their 7000 connections required over 50 person-years of labour. Despite the utility of this information, and the intrinsic glory of the knowledge, it should be noted that it most certainly can’t be said that ‘we understand the C. elegans nervous system’.

Another, very elegant technique being developed, with the capacity to map connectomes at the mesoscopic level, is the use of rabies-type viruses. These viruses can be transferred across synapses, but engineered to only do so once. Carrying fluorescent protein encoding reporter genes, they can be used to track the connecting neurons within a network. As one can engineer these viruses to carry allsorts of genetic trickery, the mapping of networks can go hand in hand with functional experimentation.

The trans-synaptic transfer of rabies viruses, along with the use of randomising recombination to create hundreds of different combinations of fluorophores in ‘brainbow’ neuronal imaging, are major inspirations underlying a new theoretical connectome mapping technique laid out by Zador et al.

The technique, termed BOINC (barcoding of individual neuronal connections) converts the mapping of the connectome from an anatomical problem into one that can be tackled by DNA sequencing. As the costs of sequencing are currently dropping through the floor, BOINC would make mapping the connectome a repeatable assay rather than a one-off mega-mission.

The method can be divided into three phases. In the first stage, each neuron is labelled with a unique DNA sequence – a barcode. The authors calculate that a random sequence of 20 nucleotides would be sufficient to individually label the entire neuronal complement of the mouse brain (1012 possible sequences to <108 neurons). Zador et al. are sketchy about the specifics of how this could be achieved; suffice to say that it’s conceptually similar to the generation of antibody diversity by recombination.

The second stage is the association of barcodes from synaptically connected neurons. This would be achieved by the single transynaptic spread of rabies-like viruses. The barcode would therefore be carried within the virus genome. In the third phase, the barcodes must be joined together. Each neuron will therefore contain its’ own barcode combined with barcodes from every cell that it synapses with. These tags would then be sequenced – yielding a connectivity matrix.

This term perhaps clarifies the biggest shortcoming of the technique; it creates a matrix, devoid of anatomical or functional detail. However, these dimensions would surely become gradually coupled with repeated experimentation.

In short, BOINC is a very clever idea. If they pull it off, it will be a great advance, allowing cheap, repeated screening of the brain’s circuitry. Obviously, it’s unlikely to be much help for studying the human brain, but providing the technical hurdles are surmountable it could revolutionize the neurobiology of model systems such as the mouse.

Zador, A.M., Dubnau, J., Oyibo, H.K., Zhan, H., Cao, G.,Peikon, I.D. (2012). Sequencing the Connectome Plos Biology, 10 (10) : 10.1371/journal.pbio.1001411

The original paper is available here

Check out some excellent blogposts by Mo Costandi at Neurophilosophy; A book review discussing the connectome concept (with special bilious email from Andrew Lumsden), and descriptions of work using Rabies viruses for mapping connectivity and functional neurobiology.

Opening up the RNA-chromatin network

In eukaryotic nuclei, DNA is coiled around histone proteins to form nucleosomes. The pattern by which nucleosomes are compacted into higher-order structures determines the accessibility of chromatin and hence it’s transcriptional activity. Many different factors, including the linker histone H1, histone modifications, chromatin remodelling enzymes and non-histone proteins play important roles in structuring chromatin. Various classes of RNA have also been implicated in regulating the higher-order structure of chromatin. Among many examples; Argonaute associated small silencing RNAs are known to sometimes exert their inhibitory effects by directing histone modifications or DNA methylation and lncRNAs have been shown to serve as cis-acting scaffolds coordinating the action of histone-modifying enzymes. It’s been known for decades that RNA makes up a proportion of chromatin, but exactly what types of RNAs and what their roles are is not yet clear. A new paper in Molecular Cell (Schubert et al.) sets out to answer these questions, characterising chromatin-associated RNAs in Drosophila and finding an important role for RNA in regulating chromatin compaction and accessibility.

Schubert et al. found that RNAs were involved in maintaining the accessibility of chromatin using an assay in which chromatin is digested by a nuclease (DNase). This digestion creates a ladder of DNA of different sizes on a gel, ranging from single nucleosomal fragments to far larger pieces. The extent of digestion is dependent on the level of chromatin condensation; the more compacted the nucleosomal structure, the more refractory it will be to DNase digestion. The researchers found that incubation of chromatin with an RNase prior to DNase treatment resulted in more compacted chromatin; DNase digestion was less efficient. Using different RNases and inhibitors they discovered that the RNA population involved was single-stranded and synthesised by RNA polymerase II.

Similar results were found when the authors used density-gradient centrifugation to isolate chromatin. They found a fraction of RNA associated with chromatin (caRNA) that when digested, resulted in chromatin becoming more compacted and shifting to higher density fractions. Interestingly, this RNA-dependent chromatin condensation effect is reversible; when the compacted chromatin was extracted and incubated with fresh cellular extracts, it reopens, again migrating in lighter fractions, re-associated with RNA. This caRNA-dependent chromatin accessibility is also dependent on chromatin-associated proteins, as it could not be rescued under denaturing conditions.

Using mass spectrometry, the authors identified 59 proteins that had lower affinity for chromatin after removal of RNA. One of these was the highly abundant chromatin decondensation factor 31 (Df31). A study that I have long intended to write about (Filion et al. 2010) used the binding of Df31 and 52 other chromatin associated proteins to determine that rather than dividing Drosophila chromatin into two types: transcriptionally repressed heterochromatin and active euchromatin, we should instead think in terms of five different classes which they colour coded. Green, blue and black chromatin are broadly transcriptionally repressed, whilst red and yellow are euchromatic and more transcriptionally active. Df31 is found bound to these red and yellow types of more open chromatin. Schubert et al found that Df31 chromatin binding is stabilised by caRNAs. Df31 binds histone H3 in the absence of RNA, but its affinity is substantially enhanced by the addition of RNA. RNAi knockdown of Df31 causes a fraction of genomic DNA to be more compacted.

Deep sequencing of the pool of caRNAs revealed that they were enriched for non-coding RNAs, especially a class termed small nucleolar RNAs (snoRNAs). snoRNAs are known to guide the modification of bases in ribosomal, transfer and messenger RNAs. They have also been implicated in RNA editing and splicing. Schubert et al. found that 30 of the 186 snoRNAs expressed in Drosophila embryos were found associated with chromatin. Using fluorescent in situ hybridisation, they showed that two of the most highly enriched snoRNAs could be visualised binding to the interbands of Drosophila polytene chromosomes (ie. euchromatin). These two snoRNAs were able to ‘rescue’ compacted chromatin in the density-gradient experiments, and were shown to directly interact with Df31.

The authors also found that extracts of human cells could rescue RNase treated compacted chromatin, and that snoRNAs are found associated with chromatin in human cells.

Schubert et al have therefore characterised a novel conserved role for snoRNAs mediating the accessibility of higher-order chromatin structures. As none of the protein components known to complex with snoRNAs in snoRNPs were identified in the mass spectrometry experiments, it appears that snoRNAs form distinct ribonucleoprotein complexes to mediate this chromatin associated role. Df31 is one such important interactor, linking chromatin-associated snoRNAs and histone proteins within nucleosomes to maintain accessibility of red and yellow chromatin. However, the knockdown of Df31 resulted in far more limited and localised chromatin compaction than RNase treatment, suggesting that the role of caRNAs goes beyond the maintenance of open euchromatin, but also to regulating heterochromatic accessibility. Seeing as 58 other chromatin proteins, as well as many other snoRNAs and other caRNAs, were implicated in RNA-associated chromatin accessibility functions, this initial study has just revealed the tip of an iceberg. Just how this RNA-chromatin network functions to maintain accessibility requires a lot of work. As with many of the best studies, the light shone serves to partially illuminate the scale of our ignorance.

Schubert T, Pusch MC, Diermeier S, Benes V, Kremmer E, Imhof A, & Längst G (2012). Df31 Protein and snoRNAs Maintain Accessible Higher-Order Structures of Chromatin. Molecular cell PMID: 23022379

Filion GJ, van Bemmel JG, Braunschweig U, Talhout W, Kind J, Ward LD, Brugman W, de Castro IJ, Kerkhoven RM, Bussemaker HJ, & van Steensel B (2010). Systematic protein location mapping reveals five principal chromatin types in Drosophila cells. Cell, 143 (2), 212-24 PMID: 20888037