Exploration of DPP-IV Inhibitory Peptide Design Rules Assisted by the Deep Learning Pipeline That Identifies the Restriction Enzyme Cutting Site.

The mining of antidiabetic dipeptidyl peptidase IV (DPP-IV) inhibitory peptides (DPP-IV-IPs) is currently a costly and laborious process. Due to the absence of rational peptide design rules, it relies on cumbersome screening of unknown enzyme hydrolysates. Here, we present an enhanced deep learning model called bidirectional encoder representation (BERT)-DPPIV, specifically designed to classify DPP-IV-IPs and explore their design rules to discover potent candidates. The end-to-end model utilizes a fine-tuned BERT architecture to extract structural/functional information from input peptides and accurately identify DPP-IV-Ips from input peptides. Experimental results in the benchmark data set showed BERT-DPPIV yielded state-of-the-art accuracy and MCC of 0.894 and 0.790, surpassing the 0.797 and 0.594 obtained by the sequence-feature model. Furthermore, we leveraged the attention mechanism to uncover that our model could recognize the restriction enzyme cutting site and specific residues that contribute to the inhibition of DPP-IV. Moreover, guided by BERT-DPPIV, proposed design rules for DPP-IV inhibitory tripeptides and pentapeptides were validated, and they can be used to screen potent DPP-IV-IPs.

Assembly and analytical validation of a metagenomic reference catalog of human gut microbiota based on co-barcoding sequencing.

Human gut microbiota is associated with human health and disease, and is known to have the second-largest genome in the human body. The microbiota genome is important for their functions and metabolites; however, accurate genomic access to the microbiota of the human gut is hindered due to the difficulty of cultivating and the shortcomings of sequencing technology. Therefore, we applied the stLFR library construction method to assemble the microbiota genomes and demonstrated that assembly property outperformed standard metagenome sequencing. Using the assembled genomes as references, SNP, INDEL, and HGT gene analyses were performed. The results demonstrated significant differences in the number of SNPs and INDELs among different individuals. The individual displayed a unique species variation spectrum, and the similarity of strains within individuals decreased over time. In addition, the coverage depth analysis of the stLFR method shows that a sequencing depth of 60X is sufficient for SNP calling. HGT analysis revealed that the genes involved in replication, recombination and repair, mobilome prophages, and transposons were the most transferred genes among different bacterial species in individuals. A preliminary framework for human gut microbiome studies was established using the stLFR library construction method.

[Functional discovery and production technology for natural bioactive peptides].

Bioactive peptides play important roles in promoting human health, such as lowering blood pressure, blood sugar and blood lipid, anti-obesity, and anti-cancer. Thus, exploring functional bioactive peptides and developing efficient production technologies are of crucial importance. Herein, we review the development of function discovery and production technology for natural bioactive peptides. Presently, the top-down and bottom-up approaches are mainly used for the function discovery and production of natural active peptides. The top-down approach includes the direct extraction and identification for functional discovery, and the direct extraction, enzymatic hydrolysis and microbial fermentation for production. The bottom-up approach includes the polypeptide modification and database mining for functional discovery, and the chemical synthesis, enzyme synthesis, recombinant expression and cell-free synthesis for production. The top-down approach is usually associated with complicated process, lower efficiency, higher cost, harder quality control, and uncertain functionality, while the bottom-up approach is more suitable for the development of peptide drugs but difficult to be used for functional foods. With the technology development of sequencing and mass spectrometry, it is easier to obtain the proteomic information of various organisms at the molecular level. Based on the proteomic information, the top-down and bottom-up approaches can be combined to overcome the disadvantages of using these two approaches alone, thus providing a new strategy for the rapid development and production of natural active peptides.

Improved sgRNA design in bacteria via genome-wide activity profiling.

CRISPR/Cas9 is a promising tool in prokaryotic genome engineering, but its success is limited by the widely varying on-target activity of single guide RNAs (sgRNAs). Based on the association of CRISPR/Cas9-induced DNA cleavage with cellular lethality, we systematically profiled sgRNA activity by co-expressing a genome-scale library (∼70 000 sgRNAs) with Cas9 or its specificity-improved mutant in Escherichia coli. Based on this large-scale dataset, we constructed a comprehensive and high-density sgRNA activity map, which enables selecting highly active sgRNAs for any locus across the genome in this model organism. We also identified 'resistant' genomic loci with respect to CRISPR/Cas9 activity, notwithstanding the highly accessible DNA in bacterial cells. Moreover, we found that previous sgRNA activity prediction models that were trained on mammalian cell datasets were inadequate when coping with our results, highlighting the key limitations and biases of previous models. We hence developed an integrated algorithm to accurately predict highly effective sgRNAs, aiming to facilitate CRISPR/Cas9-based genome engineering, screenings and antimicrobials design in bacteria. We also isolated the important sgRNA features that contribute to DNA cleavage and characterized their key differences among wild type Cas9 and its mutant, shedding light on the biophysical mechanisms of the CRISPR/Cas9 system.

Pooled CRISPR interference screening enables genome-scale functional genomics study in bacteria with superior performance.

To fully exploit the microbial genome resources, a high-throughput experimental platform is needed to associate genes with phenotypes at the genome level. We present here a novel method that enables investigation of the cellular consequences of repressing individual transcripts based on the CRISPR interference (CRISPRi) pooled screening in bacteria. We identify rules for guide RNA library design to handle the unique structure of prokaryotic genomes by tiling screening and construct an E. coli genome-scale guide RNA library (~60,000 members) accordingly. We show that CRISPRi outperforms transposon sequencing, the benchmark method in the microbial functional genomics field, when similar library sizes are used or gene length is short. This tool is also effective for mapping phenotypes to non-coding RNAs (ncRNAs), as elucidated by a comprehensive tRNA-fitness map constructed here. Our results establish CRISPRi pooled screening as a powerful tool for mapping complex prokaryotic genetic networks in a precise and high-throughput manner.

Breeding of Methanol-Tolerant Methylobacterium extorquens AM1 by Atmospheric and Room Temperature Plasma Mutagenesis Combined With Adaptive Laboratory Evolution.

Methylobacterium extorquens AM1, which can be used as a methylotrophic cell factory (MeCF) for the production of fine chemicals from methanol, is the most extensively studied model methylotrophic strain. However, its low tolerance for methanol limits the development of bioprocesses and there have been no reports of improved methanol tolerance of M. extorquens AM1. In this study, atmospheric and room temperature plasma (ARTP) mutagenesis, in combination with adaptive laboratory evolution (ALE), is used to generate a mutant with high methanol tolerance (referred to as CLY-2533). The final cell density of CLY-2533 is 7.10 times higher than that of the wild-type strain in medium containing 5% (v/v) methanol. Through comparative genomics analysis and overexpression of the exploited putative genes, seven mutated genes are identified as being closely related to the higher methanol tolerance of CLY-2533. Additionally, the mvt operon, which contains genes related to the biosynthesis of mevalonate acid (MEV), is introduced into CLY-2533. This recombinant strain shows significant improvements in both MEV production and cell growth in 5% methanol medium. These findings will be helpful in rational design of methanol-utilizing strain for an improved host platform for methanol based biomanufacturing.

Effects of Enzymatically Depolymerized Low Molecular Weight Heparins on CCl4-Induced Liver Fibrosis.

With regard to identifying the effective components of LMWH drugs curing hepatic fibrosis disease, we carried out a comparative study on the efficacy of enzymatically depolymerized LMWHs on CCl4 induced mouse liver fibrosis. The results showed that the controlled enzymatic depolymerization conditions resulted in LMWHs with significantly different activities. The LMWH product depolymerized by Heparinase I (I-11) with a Mw of 7160, exhibited a significant advantage in reducing the liver inflammation by suppressing TNF-α and IL-1ß secretion, and minimizing hepatic fibrogenesis. The products prepared by only Heparinase II (II-11), and combined Heparinase III and II (III-II-5) showed limited positive effect on hepatic inflammation and fibrosis. On the contrary, the products by combined Heparinase III and I (III-I-9, III-I-5) showed no effect or stimulation effect on the hepatic fibrogenesis. Our results provided the basis for structure-activity relationship insight for inhibition of liver fibrosis activities of LMWHs, which might have significant implications for generic anti-fibrosis disease drug development.

Biosensor-assisted transcriptional regulator engineering for Methylobacterium extorquens AM1 to improve mevalonate synthesis by increasing the acetyl-CoA supply.

Acetyl-CoA is not only an important intermediate metabolite for cells but also a significant precursor for production of industrially interesting metabolites. Methylobacterium extorquens AM1, a model strain of methylotrophic cell factories using methanol as carbon source, is of interest because it produces abundant coenzyme A compounds capable of directing to synthesis of different useful compounds from methanol. However, acetyl-CoA is not always efficiently accumulated in M. extorquens AM1, as it is located in the center of three cyclic central metabolic pathways. Here we successfully demonstrated a strategy for sensor-assisted transcriptional regulator engineering (SATRE) to control metabolic flux re-distribution to increase acetyl-CoA flux from methanol for mevalonate production in M. extorquens AM1 with introduction of mevalonate synthesis pathway. A mevalonate biosensor was constructed and we succeeded in isolating a mutated strain (Q49) with a 60% increase in mevalonate concentration (an acetyl-CoA-derived product) following sensor-based high-throughput screening of a QscR transcriptional regulator library. The mutated QscR-49 regulator (Q8*,T61S,N72Y,E160V) lost an N-terminal α-helix and underwent a change in the secondary structure of the RD-I domain at the C terminus, two regions that are related to its interaction with DNA. 13C labeling analysis revealed that acetyl-CoA flux was improved by 7% and transcriptional analysis revealed that QscR had global effects and that two key points, NADPH generation and fumC overexpression, might contribute to the carbon flux re-distribution. A fed-batch fermentation in a 5-L bioreactor for QscR-49 mutant yielded a mevalonate concentration of 2.67g/L, which was equivalent to an overall yield of 0.055mol acetyl-CoA/mol methanol, the highest yield among engineered strains of M. extorquens AM1. This work was the first attempt to regulate M. extorquens AM1 on transcriptional level and provided molecular insights into the mechanism of carbon flux regulation.

Predicting a DNA-binding protein using random forest with multiple mathematical features.

DNA-binding proteins are involved and play a crucial role in a lot of important biological processes. Hence, the identification of the DNA-binding proteins is a challenging and significant problem. In order to reveal the intrinsic information correlated to DNA-binding, nine classes of candidate features based on different mathematical fields are applied to construct the prediction model with random forest. They are fractal dimension, conjoint triad feature, Hilbert-Huang Transformation, amino acid composition, dipeptide composition, chaos game representation, and the corresponding information entropies. These mathematical expressions are evaluated with 5-fold cross validation test. The results of numerical simulations show that the mathematical features consisted of amino acid composition, fractal dimension and information entropies of amino acid and chaos game representation achieve the best performance. Its accuracy is 0.8157, and Matthew's correlation coefficient (MCC) achieves 0.5968 on the benchmark dataset from DNA-Prot. By analyzing the components of top combination of the nine candidate features, the concepts of fractal dimension and information entropy are the effective and vital features, which can provide complementary sequence-order information on the basis of amino acid composition.