Category: Uncategorized

This paper describes a method that corrects the beam induced motion (BIM) and stage drift that occurs during beam exposure in cryo-electron microscopy.

Exposure of a sample to the electron beam causes the ice-embedded particles to rotate by up to a few degrees and shift by up to a few nanometers during the image acquisition process {1}. BIM, together with the stage drift, leads to image blurring and degrading of the signal and consequently limits the maximum resolution achievable by an electron microscope.

The introduction of direct electron detectors has permitted the authors to record 16-frame movies of ice-embedded particles. Averaging of four-frame subsets and aligning images of individual particles against a reference reconstruction was carried out in order to track motion during exposure. The particle coordinates and Euler angles in each frame were derived based on those calculated for the four-frame averages. Motion compensation in each frame was found to reduce the required dataset by 10-fold and still reach similar resolutions to that obtained with conventional single-particle reconstruction.

Evolution, particularly in eukaryotes, requires the modification of genes that already exist. The ‘duplication-divergence’ model states that new genes evolve from a redundant copy of a duplicated parental gene. The copied gene is then presumed to be excluded from selection and hence prompted towards mutations that might confer a new function and an adaptive advantage. However, recent research has shown that tandem duplications are unstable and unlikely to remain long enough to acquire mutations {1,2}. 

In this paper, Näsvall and colleagues describe the innovation-amplification-divergence (IAD) model. According to this model, the parental gene of function ‘A’ first innovates by acquiring a second minor function, ‘B’. Selective pressure then favours the amplification of this gene, leading to two or more copies of the parental gene and an increased activity of ‘A’ and ‘B’. In the last stage, the copied genes are subjected to mutations that can tune their function. Eventually, the evolved genes will either be specialized in one particular function or capable of carrying out both functions at a moderately increased rate.

Elegantly, Näsvall and colleagues have studied the evolution of the histidine biosynthetic enzyme (HisA) in real time to test their model. A spontaneous hisA mutant was selected in a Salmonella enterica strain lacking an enzyme that catalyzes tryptophan synthesis (TrpF) on a medium without histidine and tryptophan. The selected hisA mutant was found to develop a low level of TrpF activity. A plasmid containing the bi-functional mutant hisA was then introduced in a HisA-/TrpF- S. enterica strain and the bacteria was grown on a medium without both histidine and tryptophan. Within a few hundred generations, the plasmid region of interest was found to be amplified; thus, increasing the expression of the bi-functional gene and the bacterial growth rate. After 3000 generations, the authors observed divergent copies of the bi-functional HisA gene. In most of the cases, divergent copies specialized to catalyze one particular reaction at a higher rate (ending up with two copies, each specialized to carry out one particular reaction). Alternatively, improved gene copies capable of catalyzing both reactions at a moderately higher rate were also found.

In this outstanding work, Karr and colleagues describe the first computational model of an organism and set up grounds for future developments in systems biology. 

One of the key challenges in modern biology is to model an organism as a whole. Doing this will enable us to predict an organism’s responses to certain conditions (e.g. exposure to drugs, gene modifications, or environmental conditions). Therefore, such a holistic approach will have a tremendous impact both in medicine and biotechnology. To model an organism requires a deep understanding not only of all pathways taking place but also of the inter-relationships between them. The work carried out by the scientific community to achieve this goal has been hampered for many years by the little knowledge extracted from experimental methods and the lack of a suitable computational approach. However, recent advances in high-throughput methods and in both software and hardware development have enabled us, for the very first time, to tackle this question with a stronger confidence. 

These authors have simulated the cell cycle of 128 Mycoplasma genitalium cells (containing 525 genes) in a typical culture environment. For this, over 900 publications with more than 1900 experimental observations were considered. The total functionality of the cell was divided into 28 different sub-models (each pertaining to a particular biological process), which were assumed to be independent between themselves on short timescales but dependent in terms of variables determined by other sub-models, based on a given timescale. 

The obtained results correlate well with other experimental data not considered for the training of the model, providing new insights into the molecular understanding of this organism and demonstrating that it is possible to classify cellular phenotypes by their underlying molecular interactions.