Adrian D. Haimovich

Genomically recoded organisms expand biological functions.

Changing the Code Easily and efficiently expanding the genetic code could provide tools to genome engineers with broad applications in medicine, energy, agriculture, and environmental safety. Lajoie et al. (p. 357) replaced all known UAG stop codons with synonymous UAA stop codons in Escherichia coli MG1655, as well as release factor 1 (RF1; terminates translation at UAG), thereby eliminating natural UAG translation function without impairing fitness. This made it possible to reassign UAG as a dedicated codon to genetically encode nonstandard amino acids while avoiding deleterious incorporation at native UAG positions. The engineered E. coli incorporated nonstandard amino acids into its proteins and showed enhanced resistance to bacteriophage T7. In a second paper, Lajoie et al. (p. 361) demonstrated the recoding of 13 codons in 42 highly expressed essential genes in E. coli. Codon usage was malleable, but synonymous codons occasionally were nonequivalent in unpredictable ways. Bacteria engineered to use nonstandard amino acids show increased resistance to bacteriophage attack. We describe the construction and characterization of a genomically recoded organism (GRO). We replaced all known UAG stop codons in Escherichia coli MG1655 with synonymous UAA codons, which permitted the deletion of release factor 1 and reassignment of UAG translation function. This GRO exhibited improved properties for incorporation of nonstandard amino acids that expand the chemical diversity of proteins in vivo. The GRO also exhibited increased resistance to T7 bacteriophage, demonstrating that new genetic codes could enable increased viral resistance.

Cell-free protein synthesis from a release factor 1 deficient Escherichia coli activates efficient and multiple site-specific nonstandard amino acid incorporation.

Site-specific incorporation of nonstandard amino acids (NSAAs) into proteins enables the creation of biopolymers, proteins, and enzymes with new chemical properties, new structures, and new functions. To achieve this, amber (TAG codon) suppression has been widely applied. However, the suppression efficiency is limited due to the competition with translation termination by release factor 1 (RF1), which leads to truncated products. Recently, we constructed a genomically recoded Escherichia coli strain lacking RF1 where 13 occurrences of the amber stop codon have been reassigned to the synonymous TAA codon (rEc.E13.ΔprfA). Here, we assessed and characterized cell-free protein synthesis (CFPS) in crude S30 cell lysates derived from this strain. We observed the synthesis of 190 ± 20 μg/mL of modified soluble superfolder green fluorescent protein (sfGFP) containing a single p-propargyloxy-l-phenylalanine (pPaF) or p-acetyl-l-phenylalanine. As compared to the parent rEc.E13 strain with RF1, this results in a modified sfGFP synthesis improvement of more than 250%. Beyond introducing a single NSAA, we further demonstrated benefits of CFPS from the RF1-deficient strains for incorporating pPaF at two- and five-sites per sfGFP protein. Finally, we compared our crude S30 extract system to the PURE translation system lacking RF1. We observed that our S30 extract based approach is more cost-effective and high yielding than the PURE translation system lacking RF1, ∼1000 times on a milligram protein produced/

Evolution of translation machinery in recoded bacteria enables multi-site incorporation of nonstandard amino acids

basis. Looking forward, using RF1-deficient strains for extract-based CFPS will aid in the synthesis of proteins and biopolymers with site-specifically incorporated NSAAs.

A flexible codon in genomically recoded Escherichia coli permits programmable protein phosphorylation

Expansion of the genetic code with nonstandard amino acids (nsAAs) has enabled biosynthesis of proteins with diverse new chemistries. However, this technology has been largely restricted to proteins containing a single or few nsAA instances. Here we describe an in vivo evolution approach in a genomically recoded Escherichia coli strain for the selection of orthogonal translation systems capable of multi-site nsAA incorporation. We evolved chromosomal aminoacyl-tRNA synthetases (aaRSs) with up to 25-fold increased protein production for p-acetyl-L-phenylalanine and p-azido-L-phenylalanine (pAzF). We also evolved aaRSs with tunable specificities for 14 nsAAs, including an enzyme that efficiently charges pAzF while excluding 237 other nsAAs. These variants enabled production of elastin-like-polypeptides with 30 nsAA residues at high yields (∼50 mg/L) and high accuracy of incorporation (>95%). This approach to aaRS evolution should accelerate and expand our ability to produce functionalized proteins and sequence-defined polymers with diverse chemistries.

Genomes by design

Biochemical investigation of protein phosphorylation events is limited by inefficient production of the phosphorylated and non-phosphorylated forms of full-length proteins. Here using a genomically recoded strain of E. coli with a flexible UAG codon we produce site-specific serine- or phosphoserine-containing proteins, with purities approaching 90%, from a single recombinant DNA. Specifically, we synthesize human MEK1 kinase with two serines or two phosphoserines, from one DNA template, and demonstrate programmable kinase activity. Programmable protein phosphorylation is poised to help reveal the structural and functional information encoded in the phosphoproteome.

Designed phosphoprotein recognition in Escherichia coli.

Next-generation DNA sequencing has revealed the complete genome sequences of numerous organisms, establishing a fundamental and growing understanding of genetic variation and phenotypic diversity. Engineering at the gene, network and whole-genome scale aims to introduce targeted genetic changes both to explore emergent phenotypes and to introduce new functionalities. Expansion of these approaches into massively parallel platforms establishes the ability to generate targeted genome modifications, elucidating causal links between genotype and phenotype, as well as the ability to design and reprogramme organisms. In this Review, we explore techniques and applications in genome engineering, outlining key advances and defining challenges.

Predicting hospital admission at emergency department triage using machine learning

Protein phosphorylation is a central biological mechanism for cellular adaptation to environmental changes. Dysregulation of phosphorylation signaling is implicated in a wide variety of diseases. Thus, the ability to detect and quantify protein phosphorylation is highly desirable for both diagnostic and research applications. Here we present a general strategy for detecting phosphopeptide–protein interactions in Escherichia coli. We first redesign a model tetratricopeptide repeat (TPR) protein to recognize phosphoserine in a sequence-specific fashion and characterize the interaction with its target phosphopeptide in vitro. We then combine in vivo site-specific incorporation of phosphoserine with split mCherry assembly to observe the designed phosphopeptide–protein interaction specificity in E. coli. This in vivo strategy for detecting and characterizing phosphopeptide–protein interactions has numerous potential applications for the study of natural interactions and the design of novel ones.

Corrigendum: Recoded organisms engineered to depend on synthetic amino acids

Objective To predict hospital admission at the time of ED triage using patient history in addition to information collected at triage. Methods This retrospective study included all adult ED visits between March 2014 and July 2017 from one academic and two community emergency rooms that resulted in either admission or discharge. A total of 972 variables were extracted per patient visit. Samples were randomly partitioned into training (80%), validation (10%), and test (10%) sets. We trained a series of nine binary classifiers using logistic regression (LR), gradient boosting (XGBoost), and deep neural networks (DNN) on three dataset types: one using only triage information, one using only patient history, and one using the full set of variables. Next, we tested the potential benefit of additional training samples by training models on increasing fractions of our data. Lastly, variables of importance were identified using information gain as a metric to create a low-dimensional model. Results A total of 560,486 patient visits were included in the study, with an overall admission risk of 29.7%. Models trained on triage information yielded a test AUC of 0.87 for LR (95% CI 0.86–0.87), 0.87 for XGBoost (95% CI 0.87–0.88) and 0.87 for DNN (95% CI 0.87–0.88). Models trained on patient history yielded an AUC of 0.86 for LR (95% CI 0.86–0.87), 0.87 for XGBoost (95% CI 0.87–0.87) and 0.87 for DNN (95% CI 0.87–0.88). Models trained on the full set of variables yielded an AUC of 0.91 for LR (95% CI 0.91–0.91), 0.92 for XGBoost (95% CI 0.92–0.93) and 0.92 for DNN (95% CI 0.92–0.92). All algorithms reached maximum performance at 50% of the training set or less. A low-dimensional XGBoost model built on ESI level, outpatient medication counts, demographics, and hospital usage statistics yielded an AUC of 0.91 (95% CI 0.91–0.91). Conclusion Machine learning can robustly predict hospital admission using triage information and patient history. The addition of historical information improves predictive performance significantly compared to using triage information alone, highlighting the need to incorporate these variables into prediction models.

Recoded organisms engineered to depend on synthetic amino acids

This corrects the article DOI: 10.1038/nature14095