The stepwise reduction of dihydrofolate to tetrahydrofolate entails significant conformational changes of dihydrofolate reductase (DHFR). Binary and ternary complexes of DHFR containing cofactor NADPH, inhibitor methotrexate (MTX), or both NADPH and MTX were characterized by 193 nm ultraviolet photodissociation (UVPD) mass spectrometry. UVPD yielded over 80% sequence coverage of DHFR and resulted in production of fragment ions that revealed the interactions between DHFR and each ligand. UVPD of the binary DHFR·NADPH and DHFR·MTX complexes led to an unprecedented number of fragment ions containing either an N- or C-terminal protein fragment still bound to the ligand via retention of noncovalent interactions. In addition, holo-fragments retaining both ligands were observed upon UVPD of the ternary DHFR·NADPH·MTX complex. The combination of extensive holo and apo fragment ions allowed the locations of the NADPH and MTX ligands to be mapped, with NADPH associated with the adenosine binding domain of DHFR and MTX interacting with the loop domain. These findings are consistent with previous crystallographic evidence. Comparison of the backbone cleavage propensities for apo DHFR and its holo counterparts revealed significant variations in UVPD fragmentation in the regions expected to experience conformational changes upon binding NADPH, MTX, or both ligands. In particular, the subdomain rotation and loop movements, which are believed to occur upon formation of the transition state of the ternary complex, are reflected in the UVPD mass spectra. The UVPD spectra indicate enhanced backbone cleavages in regions that become more flexible or show suppressed backbone cleavages for those regions either shielded by the ligand or involved in new intramolecular interactions. This study corroborates the versatility of 193 nm UVPD mass spectrometry as a sensitive technique to track enzymatic cycles that involve conformational rearrangements.

Despite the promise of deep learning accelerated protein engineering, examples of such improved proteins are scarce. Here we report that a 3D convolutional neural network trained to associate amino acids with neighboring chemical microenvironments can guide identification of novel gain-of-function mutations that are not predicted by energetics-based approaches. Amalgamation of these mutations improved protein function in vivo across three diverse proteins by at least 5-fold. Furthermore, this model provides a means to interrogate the chemical space within protein microenvironments and identify specific chemical interactions that contribute to the gain-of-function phenotypes resulting from individual mutations.

Bacterial selenocysteine incorporation occurs in response to opal stop codons and is dependent on the presence of a selenocysteine insertion sequence (SECIS) element, which recruits the selenocysteine specific elongation factor and tRNASec needed to reassign the UGA codon. The SECIS element is a stem-loop RNA structure immediately following the UGA codon and forms part of the coding sequence in bacterial selenoproteins. Although the site specific incorporation of selenocysteine is of great interest for protein engineering, the sequence constraints imposed by the adjoining SECIS element severely limit its use. We have evolved an E. coli tRNASec that is compatible with the canonical translation machinery and can suppress amber stop codons to incorporate selenocysteine with high efficiency. This evolved tRNASec allows the production of new recombinant selenoproteins containing structural motifs such as selenyl-sulfhydryl and diselenide bonds.

The cytosolic and mitochondrial thioredoxin reductases (TrxR1 and TrxR2) and thioredoxins (Trx1 and Trx2) are key components of the mammalian thioredoxin system, which is important for antioxidant defense and redox regulation of cell function. TrxR1 and TrxR2 are selenoproteins generally considered to have comparable properties, but to be functionally separated by their different compartments. To compare their properties we expressed recombinant human TrxR1 and TrxR2 and determined their substrate specificities and inhibition by metal compounds. TrxR2 preferred its endogenous substrate Trx2 over Trx1, whereas TrxR1 efficiently reduced both Trx1 and Trx2. TrxR2 displayed strikingly lower activity with dithionitrobenzoic acid (DTNB), lipoamide, and the quinone substrate juglone compared to TrxR1, and TrxR2 could not reduce lipoic acid. However, Sec-deficient two-amino-acid-truncated TrxR2 was almost as efficient as full-length TrxR2 in the reduction of DTNB. We found that the gold(I) compound auranofin efficiently inhibited both full-length TrxR1 and TrxR2 and truncated TrxR2. In contrast, some newly synthesized gold(I) compounds and cisplatin inhibited only full-length TrxR1 or TrxR2 and not truncated TrxR2. Surprisingly, one gold(I) compound, [Au(d2pype)(2)]Cl, was a better inhibitor of TrxR1, whereas another, [(iPr(2)Im)(2)Au]Cl, mainly inhibited TrxR2. These compounds also inhibited TrxR activity in the cytoplasm and mitochondria of cells, but their cytotoxicity was not always dependent on the proapoptotic proteins Bax and Bak. In conclusion, this study reveals significant differences between human TrxR1 and TrxR2 in substrate specificity and metal compound inhibition in vitro and in cells, which may be exploited for development of specific TrxR1- or TrxR2-targeting drugs.

Engineered orthogonal translation systems have greatly enabled the expansion of the genetic code using noncanonical amino acids (NCAAs). However, the impact of NCAAs on organismal evolution remains unclear, in part because it is difficult to force the adoption of new genetic codes in organisms. By reengineering TEM-1 β-lactamase to be dependent on a NCAA, we maintained bacterial NCAA dependence for hundreds of generations without escape.

Native mass spectrometry, size exclusion chromatography, and kinetic assays were employed to study trimethoprim resistance in E. coli caused by mutations P21L and W30R of dihydrofolate reductase.

Incorporation of the rare amino acid selenocysteine to form diselenide bonds can improve stability and function of synthetic peptide therapeutics. However, application of this approach to recombinant proteins has been hampered by heterogeneous incorporation, low selenoprotein yields, and poor fitness of bacterial producer strains. We report the evolution of recoded E. coli strains with improved fitness that are superior hosts for recombinant selenoprotein production. We apply an engineered β-lactamase containing an essential diselenide bond to enforce selenocysteine dependence during continuous evolution of recoded E. coli strains. Evolved strains maintain an expanded genetic code indefinitely. We engineer a fluorescent reporter to quantify selenocysteine incorporation in vivo and show complete decoding of UAG codons as selenocysteine. Replacement of native, labile disulfide bonds in antibody fragments with diselenide bonds vastly improves resistance to reducing conditions. Highly seleno-competent bacterial strains enable industrial-scale selenoprotein expression and unique diselenide architecture, advancing our ability to customize the selenoproteome.

We developed a new genetic selection approach to screen for mutations that can alter the efficiency of selenocysteine incorporation. We identified mutations in 16S rRNA that increase or decrease the efficiency of selenocysteine incorporation in Escherichia coli without influencing the efficiency or fidelity of canonical translation. Engineered ribosomes with improved selenocysteine incorporation provide valuable tools for synthetic biology and biotechnology.