The editing reactions catalyzed by aminoacyl-tRNA synthetases are critical for the faithful protein synthesis by correcting misactivated amino acids and misaminoacylated tRNAs. We report that the isolated editing domain of leucyl-tRNA synthetase from the deep-rooted bacterium Aquifex aeolicus (alphabeta-LeuRS) catalyzes the hydrolytic editing of both mischarged tRNA(Leu) and minihelix(Leu). Within the domain, we have identified a crucial 20-amino-acid peptide that confers editing capacity when transplanted into the inactive Escherichia coli LeuRS editing domain. Likewise, fusion of the beta-subunit of alphabeta-LeuRS to the E. coli editing domain activates its editing function. These results suggest that alphabeta-LeuRS still carries the basic features from a primitive synthetase molecule. It has a remarkable capacity to transfer autonomous active modules, which is consistent with the idea that modern synthetases arose after exchange of small idiosyncratic domains. It also has a unique alphabeta-heterodimeric structure with separated catalytic and tRNA-binding sites. Such an organization supports the tRNA/synthetase coevolution theory that predicts sequential addition of tRNA and synthetase domains.

In a hyperthermophilic bacterium, Aquifex aeolicus, leucyl-tRNA synthetase (LeuRS) consists of two nonidentical polypeptide subunits (␣ and ␤), different from the canonical LeuRS, which has a single polypeptide chain. By PCR, using genome DNA of A. aeolicus as a template, genes encoding the ␣ and ␤ subunits were amplified and cloned in Escherichia coli. The ␣ subunit could not be expressed stably in vivo, whereas the ␤ subunit was overproduced and purified by a simple procedure. The ␤ subunit was inactive in catalysis but was able to bind tRNA Leu . Interestingly, the heterodimer ␣␤-LeuRS could be overproduced in E. coli cells containing both genes and was purified to 95% homogeneity as a hybrid dimer. The kinetics of A. aeolicus LeuRS in presteady and steady states and cross-recognition of LeuRS and tRNA Leu from A. aeolicus and E. coli were studied. Magnesium concentration, pH value, and temperature aminoacylation optima were determined to be 12 mM, 7.8, and 70°C, respectively. Under optimal conditions, A. aeolicus ␣␤-LeuRS is stable up to 65°C.

Label contamination attack (LCA) is an important type of data poisoning attack where an attacker manipulates the labels of training data to make the learned model beneficial to him. Existing work on LCA assumes that the attacker has full knowledge of the victim learning model, whereas the victim model is usually a black-box to the attacker. In this paper, we develop a Projected Gradient Ascent (PGA) algorithm to compute LCAs on a family of empirical risk minimizations and show that an attack on one victim model can also be effective on other victim models. This makes it possible that the attacker designs an attack against a substitute model and transfers it to a black-box victim model. Based on the observation of the transferability, we develop a defense algorithm to identify the data points that are most likely to be attacked. Empirical studies show that PGA significantly outperforms existing baselines and linear learning models are better substitute models than nonlinear ones.