Deep-Learning Approach to First-Principles Transport Simulations.

Large-scale first-principles transport calculations, while essential for device modeling, remain computationally demanding. To overcome this bottle neck, we combine first-principles transport calculations with machine learning-based nonlinear regression. We calculate the electronic conductance through first-principles based nonequilibrium Green's function techniques for small systems and map the transport properties onto local properties using local descriptors. We show that using the local descriptor as input features for deep learning-based nonlinear regression allows us to build a robust neural network that can predict the conductance of large systems beyond that of the current state-of-the-art first-principles calculation algorithms. Our protocol is applied to alkali metal nanowires, i.e., potassium, which have unique geometrical and electronic properties and hence nontrivial transport properties. We demonstrate that within our approach we can achieve qualitative agreement with experiment at a fraction of the computational effort as compared to the direct calculation of the transport properties using conventional first-principles methods.

DNA Origami Scaffolds as Templates for Functional Tetrameric Kir3 K+ Channels.

In native systems, scaffolding proteins play important roles in assembling proteins into complexes to transduce signals. This concept is yet to be applied to the assembly of functional transmembrane protein complexes in artificial systems. To address this issue, DNA origami has the potential to serve as scaffolds that arrange proteins at specific positions in complexes. Herein, we report that Kir3 K+ channel proteins are assembled through zinc-finger protein (ZFP)-adaptors at specific locations on DNA origami scaffolds. Specific binding of the ZFP-fused Kir3 channels and ZFP-based adaptors on DNA origami were confirmed by atomic force microscopy and gel electrophoresis. Furthermore, the DNA origami with ZFP binding sites nearly tripled the K+ channel current activity elicited by heterotetrameric Kir3 channels in HEK293T cells. Thus, our method provides a useful template to control the oligomerization states of membrane protein complexes in vitro and in living cells.

3D structure determination of protein using TEM single particle analysis.

Proteins play important roles in cell functions such as enzymes, cell trafficking, neurotransmission, muscle contraction and hormone secretion. However, some proteins are very difficult to be crystallized and their structures are undetermined. Several techniques have been developed to elucidate the structure of macromolecules; X-ray or electron crystallography, nuclear magnetic resonance spectroscopy, and high-resolution electron microscopy. Among them, electron microscopy based single particle reconstruction (SPA) technique is a computer-aided structure determination method. This method reconstructs the 3D structure from projection images of dispersed protein. A large number of two-dimensional particle images are picked up from EM films, aligned and classified to generate 2D averages, and used to reconstruct the 3D structure by assigning the Euler angle of each 2D average. Due to the necessity of elaborate collaboration between the classical biology and the innovative information technology including parallel computing, scientists needed to break unseen barriers to get a start of this analysis. However, recent progresses in electron microscopes, mathematical algorithms, and computational abilities greatly reduced the height of barriers and expanded targets that are considered to be primarily addressable using single particle analysis. Membrane proteins are one of these targets to which the single particle analysis is successfully applied for the understanding of their 3D structures. For this purpose, we have developed various SPA methods [1-5] and applied them to different proteins [6-8].Here, we introduce reconstructed proteins, and discuss the availability of this technique. The intramembrane-cleaving proteases (I-CLiPs) that sever the transmembrane domains of their substrates have been identified in a range of organisms and play a variety of roles in biological conditions. I-CLiPs have been classified into three groups: serine-, aspartyl- and metalloprotease-type. Signal peptide peptidase (SPP) is an atypical aspartic protease that hydrolyzes peptide bonds within the transmembrane domain of substrates and is implicated in several biological and pathological functions. The structure of human SPP was determined by SPA at a resolution of 22 Å [8]. SPP forms a slender, bullet-shaped homotetramer with dimensions of 85 x 85 x 130 Å. The SPP complex has four concaves on the rhombus-like sides, connected to a large chamber inside the molecule. For the tetrameric assembly, the N-terminal region of SPP was found to be sufficient. Moreover, when N-terminal region was overexpressed, the formation of the endogenous SPP tetramer was inhibited, which suppressed the proteolytic activity within cells. From these data, the N-terminal region is considered to work as the structural scaffold.Transmembrane (TM) translocation of newly synthesized secretion proteins and membrane proteins are carried out by a Sec translocon protein complex. The polypeptide-conducting pore is formed by the SecYEG-SecA complex in bacteria, and the membrane protein SecDF is necessary for the efficient transport of proteins. However the molecular mechanism how SecDF realized efficient transport is not clear. A previous X-ray structural study of the whole protein and subdomain suggest that SecDF has at least two conformational variants, which could reflect molecular dynamics of this protein. To confirm this hypothesis, we analyzed the 3D structure of SecDF using dark field STEM electron tomography and single particle reconstruction. We determined two different whole SecDF protein structures which well explains the X-ray data. From these data, we would like to propose the possible molecular mechanism of SecDF during polypeptide translocation.

Immuno-electron microscopy of primary cell cultures from genetically modified animals in liquid by atmospheric scanning electron microscopy.

High-throughput immuno-electron microscopy is required to capture the protein-protein interactions realizing physiological functions. Atmospheric scanning electron microscopy (ASEM) allows in situ correlative light and electron microscopy of samples in liquid in an open atmospheric environment. Cells are cultured in a few milliliters of medium directly in the ASEM dish, which can be coated and transferred to an incubator as required. Here, cells were imaged by optical or fluorescence microscopy, and at high resolution by gold-labeled immuno-ASEM, sometimes with additional metal staining. Axonal partitioning of neurons was correlated with specific cytoskeletal structures, including microtubules, using primary-culture neurons from wild type Drosophila, and the involvement of ankyrin in the formation of the intra-axonal segmentation boundary was studied using neurons from an ankyrin-deficient mutant. Rubella virus replication producing anti-double-stranded RNA was captured at the host cell's plasma membrane. Fas receptosome formation was associated with clathrin internalization near the surface of primitive endoderm cells. Positively charged Nanogold clearly revealed the cell outlines of primitive endoderm cells, and the cell division of lactic acid bacteria. Based on these experiments, ASEM promises to allow the study of protein interactions in various complexes in a natural environment of aqueous liquid in the near future.

Conformational variation of the translocon enhancing chaperone SecDF.

The Sec translocon facilitates transportation of newly synthesized polypeptides from the cytoplasm to the lumen/periplasm across the phospholipid membrane. Although the polypeptide-conducting machinery is formed by the SecYEG-SecA complex in bacteria, its transportation efficiency is markedly enhanced by SecDF. A previous study suggested that SecDF assumes at least two conformations differing by a 120° rotation in the spatial orientation of the P1 head subdomain to the rigid base, and that the conformational dynamics plays a critical role in polypeptide translocation. Here we addressed this hypothesis by analyzing the 3D structure of SecDF using electron tomography and single particle reconstruction. Reconstruction of wt SecDF showed two major conformations; one resembles the crystal structure of full-length SecDF (F-form structure), while the other is similar to the hypothetical structural variant based on the crystal structure of the isolated P1 domain (I-form structure). The transmembrane domain of the I-form structure has a scissor like cleft open to the periplasmic side. We also report the structure of a double cysteine mutant designed to constrain SecDF to the I-form. This reconstruction has a protrusion at the periplasmic end that nicely fits the orientation of P1 in the I-from. These results provide firm evidence for the occurrence of the I-form in solution and support the proposed F- to I-transition of wt SecDF during polypeptide translocation.

Expression, purification, crystallization and preliminary crystallographic analysis of hepatitis B virus core protein dimerized via a peptide linker containing an EGFP insertion.

Virus-like particles (VLPs) have many potentially useful applications. The core proteins of human hepatitis B virus self-assemble into icosahedral VLPs. As previously reported, core protein dimers (CPDs), produced by connecting two core proteins via a peptide linker, can also assemble into VLPs. CPDs in which heterologous proteins were connected to the C-terminus (CPD1) were found to rearrange into symmetrical octahedra during crystallization. In this study, a heterologous protein was inserted into the peptide linker of the CPD (CPD2). CPD2 was expressed in Escherichia coli, assembled into VLPs, purified and crystallized. A single crystal diffracted to 2.8 Å resolution and belonged to the cubic space group F432, with unit-cell parameters a = b = c = 218.6 Å. Single-crystal analysis showed that CPD1 and CPD2 rearranged into the same octahedral organization in a crystallization solution.

Expression, purification, crystallization and preliminary X-ray crystallographic studies of hepatitis B virus core fusion protein corresponding to octahedral particles.

Recombinant hepatitis B virus core proteins dimerize to form building blocks that are capable of self-assembly into a capsid. A core capsid protein dimer (CPD) linked to a green fluorescent protein variant, EGFP, at the C-terminus has been designed. The recombinant fusion CPD was expressed in Escherichia coli, assembled into virus-like particles (VLPs), purified and crystallized. The single crystal diffracted to 2.15 Å resolution and belonged to the cubic space group F432, with unit-cell parameters a = b = c = 219.7 Å. The fusion proteins assembled into icosahedral VLPs in aqueous solution, but were rearranged into octahedral symmetry through the crystal-packing process under the crystallization conditions.

Multi-reference-based multiple alignment statistics enables accurate protein-particle pickup from noisy images.

Data mining from noisy data/images is one of the most important themes in modern science and technology. Statistical image processing is a promising technique for analysing such data. Automation of particle pickup from noisy electron micrographs is essential, especially when improvement of the resolution of single particle analysis requires a huge number of particle images. For such a purpose, reference-based matching using primary three-dimensional (3D) model projections is mainly adopted. In the matching, however, the highest peaks of the correlation may not accurately indicate particles when the image is very noisy. In contrast, the density and the heights of the peaks should reflect the probability distribution of the particles. To statistically determine the particle positions from the peak distributions, we have developed a density-based peak search followed by a peak selection based on average peak height, using multi-reference alignment (MRA). Its extension, using multi-reference multiple alignment (MRMA), was found to enable particle pickup at higher accuracy even from extremely noisy images with a signal-to-noise ratio of 0.001. We refer to these new methods as stochastic pickup with MRA (MRA-StoPICK) or with MRMA (MRMA-StoPICK). MRMA-StoPICK has a higher pickup accuracy and furthermore, is almost independent of parameter settings. They were successfully applied to cryo-electron micrographs of Rice dwarf virus. Because current computational resources and parallel data processing environments allow somewhat CPU-intensive MRA-StoPICK and MRMA-StoPICK to be performed in a short period, these methods are expected to allow high-resolution analysis of the 3D structure of particles.

Single particle reconstruction of membrane proteins: a tool for understanding the 3D structure of disease-related macromolecules.

Membrane proteins play important roles in cell functions such as neurotransmission, muscle contraction, and hormone secretion, but their structures are mostly undetermined. Several techniques have been developed to elucidate the structure of macromolecules; X-ray or electron crystallography, nuclear magnetic resonance spectroscopy, and high-resolution electron microscopy. Electron microscopy-based single particle reconstruction, a computer-aided structure determination method, reconstructs a three-dimensional (3D) structure from projections of monodispersed protein. A large number of particle images are picked up from EM films, aligned and classified to generate two-dimensional (2D) averages, and, using the Euler angle of each 2D average, reconstructed into a 3D structure. This method is challenging due to the necessity for close collaboration between classical biochemistry and innovative information technology, including parallel computing. However, recent progress in electron microscopy, mathematical algorithms, and computational ability has greatly increased the subjects that are considered to be primarily addressable using single particle reconstruction. Membrane proteins are one of these targets to which the single particle reconstruction is successfully applied for understanding of their structures. In this paper, we will introduce recently reconstructed channel-related proteins and discuss the applicability of this technique in understanding molecular structures and their roles in pathology.

A statistically harmonized alignment-classification in image space enables accurate and robust alignment of noisy images in single particle analysis.

In determining the three-dimensional (3D) structure of macromolecular assemblies in single particle analysis, a large representative dataset of two-dimensional (2D) average images from huge number of raw images is a key for high resolution. Because alignments prior to averaging are computationally intensive, currently available multireference alignment (MRA) software does not survey every possible alignment. This leads to misaligned images, creating blurred averages and reducing the quality of the final 3D reconstruction. We present a new method, in which multireference alignment is harmonized with classification (multireference multiple alignment: MRMA). This method enables a statistical comparison of multiple alignment peaks, reflecting the similarities between each raw image and a set of reference images. Among the selected alignment candidates for each raw image, misaligned images are statistically excluded, based on the principle that aligned raw images of similar projections have a dense distribution around the correctly aligned coordinates in image space. This newly developed method was examined for accuracy and speed using model image sets with various signal-to-noise ratios, and with electron microscope images of the Transient Receptor Potential C3 and the sodium channel. In every data set, the newly developed method outperformed conventional methods in robustness against noise and in speed, creating 2D average images of higher quality. This statistically harmonized alignment-classification combination should greatly improve the quality of single particle analysis.