Field of Science

Finding a new translation factor, and verifying it with help from my experimental friends

Joy of joys my most recent paper has just been published in Nucleic Acids Research!

Evolutionary and genetic analyses of mitochondrial translation initiation factors identify the missing mitochondrial IF3 in S. cerevisiae.

I'll give a little bit of background to the paper. I actually alluded to it last year when we were in the early stages of writing up.

It started with these papers by Gaur et al and Yassin et al, which claim that an insertion in mitochondrial initiation factor mIF2 evolved to compensate for a total lack of IF1 in mitochondria, which is universal in bacteria. This idea is really cute, and in fact they found that bovine mIF2 compensates for a lack of IF2+IF1 in E. coli. In addition, cryo-EM structures suggest the insertion occupies an overlapping site of the ribosome as IF1 does. However, I made an alignment of mIF2 from a broad distribution of eukaryotes and found there's a significant flaw with this hypothesis: mIF1 is absent in all eukaryotes, but the conserved insertion is only present in vertebrates. So somehow most eukaryotes through most of their evolutionary history have managed just fine without IF1 or the mIF2 insertion. This is a really nice example of how you really should do the comparative genomics of you want to make such sweeping claims as X evolved to replace Y. Here's the pic I made for my older blog post:

Just replace "protein 1" with IF2 and "protein 2" with IF1. Simple! The point is, if you only look at humans and bacteria, you may jump to a conclusion that isn't supported when you sample broadly across taxa.

My partner in life and crime, Vasili and I were thinking of writing up a teeny tiny paper about this, but then I started checking out more mitochondrial initiation factors...

Most eukaryotes carry have an orthologue of mitochondrial mIF3, but a homologue had never been found in Saccharomyces cerevisiae. Considering how important IF3 is for bacterial translation, and also how human mIF3 is associated with Parkinson's disease, it was a real bummer that this wasn't present in the yeast mitochondrial system. Well, actually, people just weren't looking hard enough. mIF3 is a small protein with a very biased amino acid content, so if just searching with BlastP, you only pick up a few of the most closely related sequences. I tried the more sensitive PSI-Blast and found an S. cerevisiae homolog called Aim23p. Aim23p was known to be a mitochondrial protein, but its function had not been predicted. Phylogenetic analysis confirmed Aim23p is the orthologue of mIF3. Cool! So our teeny tiny paper suddenly got more substantial.

The two domains of mIF3. The yellow sites are those that are strongly conserved between mIF3 and Aim23p. They are mainly internal, probably important for stabilising the structure. The outer faces are more variable, possibly reflecting lineage-specific evolution of inter-molecular interactions.

But the really cool stuff happened when we got our friends and collaborators from the lab of Piotr Kamenski in Moscow on board. Piotr's group showed that a knock-out of Aim23p can be complemented by Schizosaccharomyces pombe mIF3, strongly suggesting that Aim23 is the functional equivalent of mIF3 as well as the evolutionary orthologue.

I'm really happy about the results of this paper. It's a great feeling to predict something in silico that gets verified in vivo. Our next collaborative project with Piotr's group is a longer shot, but also bloody exciting, so fingers crossed!

Atkinson, G., Kuzmenko, A., Kamenski, P., Vysokikh, M., Lakunina, V., Tankov, S., Smirnova, E., Soosaar, A., Tenson, T., & Hauryliuk, V. (2012). Evolutionary and genetic analyses of mitochondrial translation initiation factors identify the missing mitochondrial IF3 in S. cerevisiae Nucleic Acids Research DOI: 10.1093/nar/gks272

Switches and latches on the ribosome

This post was chosen as an Editor's Selection for
Call me a geek, but I'm a big fan of translational GTPases (trGTPases). They're essential factors for translation by the ribosome, highly conserved in all life and they're ancient. Really ancient. There were at least three (EF-Tu, IF2 and EF-G) in the ancestor of all life on earth. These have subsequently diversified through gene duplication and subfunctionalisation into the multiple trGTPase subfamilies that are known today.

Since a lot of my research is on trGTPase evolution,  I was really interested to see this paper today in Nature Structural and Molecular Biology: A conserved proline switch on the ribosome facilitates the recruitment and binding of trGTPases. 

In this paper, Wang et al have identified a universally conserved residue, Pro22 of ribosomal protein L11 that switches conformation upon trGTPase EF-G binding, controlling the L11 and L12 protein interactions required for each peptide elongation cycle. They find that EF-G carries peptidyl-prolyl cis-trans isomerase (PPIase) activity, driving conformational switching of two adjacent prolines (PS22) from a trans to cis orientation. 

Most exciting for me is their claim that all known universal trGTPases contain an active PPIase center,  and therefore that the cis-trans isomerization of PS22 is a universal event required for efficient turnover of trGTPases throughout the translation process. This is pretty cool. Hang on though... structures show that the potential PPIase motif is located in the cleft between domains G and V of EF-G, but the other trGTPases they mention in the paper (IF2, EF-Tu, EF-G, EF4 (LepA) and RF3) don't carry a domain homologous to EF-G's C-terminal domain V, so how does that work? Their structural figures of RF3, EF-Tu and LepA show their predictions of where the PPIase centre might be in these proteins. In each case, the centre is formed by residues of the GTPase (G) domain that correspond to the EF-G PPIase sites, plus nearby residues in the non-homologous C-terminal domains of these various trGTPases. So we would predict that if this is a universal mechanism, at least those residues of the G domain that are involved should be universally conserved. Unfortunately, an alignment of this region isn't presented in the paper, but fortunately I have one up my sleeve:

The region of the G domain containing sites proposed to be acting as a PPIase (yellow highlighting). These are just upstream of the G4 nucleotide binding motif (highlighted in turquoise). These are consensus sequences: the residue shown for each trGTPase in each position is the most common amino acid found there across all sequences of the subfamily. In essence, each sequence is an average of hundreds of sequences.
The G domain is generally very well conserved across its length, with patches of universal conservation. However, the potential PPIase sites (yellow) are not universally conserved. The variability of these sites contrasts sharply with the nearby conserved G4 motif (one of the five motifs of the trGTPases - and in fact all GTPases - that coordinate the GTP/GDP nucleotides). It's disappointing, but this pattern of conservation doesn't really bear the hallmark of a universal mechanism. Indeed, Wang et al find that mutating the PPIase motif of EF-G has only a modest effect: a single turn of GTP hydrolysis was unaffected, but the multiple- turnover rates were inhibited by 15–40%. Similarly, mutation of PS22 did not completely abolish translation.

In fact, L11 is actually one of the few ribosomal proteins that is not essential for life, which makes L11 knock out strains useful molecular biological tools (for example this paper). Therefore, the mechanistic details of this protein's role in translation, although very interesting of course, do not translate to an understanding of the core principles of how trGTPase work on the ribosome.  

Are we really looking at something as concrete as a conserved "switch-and-latch" mechanism, or is PS22 just a trGTPase binding site with integral flexibility? Anyway, it's all interesting additional details of how trGTPases interact with the ribosome, and another plus: it got me blogging again after a long hiatus!

Wang L, Yang F, Zhang D, Chen Z, Xu RM, Nierhaus KH, Gong W, & Qin Y (2012). A conserved proline switch on the ribosome facilitates the recruitment and binding of trGTPases. Nature Structural & Molecular Biology PMID: 22407015

PhD position available in Molecular Evolution!

**** EDIT The position has now been filled and I'm no longer taking applications ****

Having got my Estonian Science Foundation grant funded recently, I have an open PhD position available! See below.

We are seeking a highly motivated PhD candidate to be supervised by Dr Gemma Atkinson within the group of Prof Tanel Tenson in the Institute of Technology, University of Tartu, Estonia.

Dr Atkinson’s research addresses protein functional evolution, using bioinformatic approaches and primarily focusing on the ancient families of proteins involved in translation of mRNA to protein. Members of these families are often essential for life and predate the last common ancestor of all life on earth. Thus by studying these proteins we can gain understanding of the fundamental processes of life, and how these processes have evolved over billions of years.

The PhD project will take advantage of the thousands of whole genome sequences now available for the study of evolution of protein families from the origin of life to the present day. Work will involve sensitive sequence searching to identify the presence and absence of particular proteins across genomes, phylogenetic analyses to reconstruct their emergence and evolution, and sequence analyses to link domain- and site-specific patterns of amino acid substitution with molecular function. Specifically, the proposed PhD project will target the ABC superfamily of ATP-binding enzymes found in all domains of life.  This superfamily comprises enigmatic proteins of diverse, and often unknown functions. Several ABC enzymes have recently been found to have important roles in regulation of translation such as ribosome recycling protein Rli1/ABCE1, yeast-specific elongation factor eEF3 and starvation response enzymes Gcn1 and Gcn20.

From the results of the PhD, it is expected that enzymes with novel roles in protein synthesis will come to light as interesting targets for subsequent experimental study. Dr Atkinson collaborates with the lab Dr Vasili Hauryliuk, also in Prof Tenson’s group, for biochemical and genetic validation of in silico results. If the candidate so wishes, there is an opportunity to gain practical lab experience in Dr Hauryliuk’s lab.

The candidate should have:
  • a Masters degree in a biological or computational discipline
  • a strong interest in, and enthusiasm for molecular evolution
  • familiarity with basic sequence and phylogenetic analyses
  • experience in using a programming language such as Python, Perl, Java etc
  • fluency in spoken and written English

Estonia has a rich culture and beautiful natural environment, with unspoiled forests, meadows and coastlines. Enjoying warm summers and cold winters, the historical city of Tartu is the intellectual capital of Estonia, and its university is the leading research and development institution in the country. The Institute of Technology is a lively, modern centre for biological and technological research.

The PhD will be funded by a monthly stipend, with additional monies available for regular attendance at international conferences and workshops, and for visiting labs abroad. Information on funding is available by request.

Applications should contain:
  • a full CV with detailed description of previous relevant experience
  • a statement of academic interests
  • an electronic version of the Masters thesis
  • the names and contact details of at least 2 referees

The candidate is expected to start at the latest September 2012. Please send applications and informal enquiries to

Gemma Atkinson
University of Tartu,
Institute of Technology
Nooruse  1, 50411 Tartu, Estonia

More information about the research of Dr Atkinson can be found here: