A kinase-cGAS cascade to synthesize a therapeutic STING activator.

The introduction of molecular complexity in an atom- and step-efficient manner remains an outstanding goal in modern synthetic chemistry. Artificial biosynthetic pathways are uniquely able to address this challenge by using enzymes to carry out multiple synthetic steps simultaneously or in a one-pot sequence1-3. Conducting biosynthesis ex vivo further broadens its applicability by avoiding cross-talk with cellular metabolism and enabling the redesign of key biosynthetic pathways through the use of non-natural cofactors and synthetic reagents4,5. Here we describe the discovery and construction of an enzymatic cascade to MK-1454, a highly potent stimulator of interferon genes (STING) activator under study as an immuno-oncology therapeutic6,7 (ClinicalTrials.gov study NCT04220866 ). From two non-natural nucleotide monothiophosphates, MK-1454 is assembled diastereoselectively in a one-pot cascade, in which two thiotriphosphate nucleotides are simultaneously generated biocatalytically, followed by coupling and cyclization catalysed by an engineered animal cyclic guanosine-adenosine synthase (cGAS). For the thiotriphosphate synthesis, three kinase enzymes were engineered to develop a non-natural cofactor recycling system in which one thiotriphosphate serves as a cofactor in its own synthesis. This study demonstrates the substantial capacity that currently exists to use biosynthetic approaches to discover and manufacture complex, non-natural molecules.

Structural Basis for the Effects of Phenylalanine on Tuning the Reduction Potential of Type 1 Copper in Azurin.

In order to investigate the effects of the secondary coordination sphere in fine-tuning redox potentials (E°') of type 1 blue copper (T1Cu) in cupredoxins, we have introduced M13F, M44F, and G116F mutations both individually and in combination in the secondary coordination sphere of the T1Cu center of azurin (Az) from Pseudomonas aeruginosa. These variants were found to differentially influence the E°' of T1Cu, with M13F Az decreasing E°', M44F Az increasing E°', and G116F Az showing a negligible effect. In addition, combining the M13F and M44F mutations increases E°' by 26 mV relative to WT-Az, which is very close to the combined effect of E°' by each mutation. Furthermore, combining G116F with either M13F or M44F mutation resulted in negative and positive cooperative effects, respectively. Crystal structures of M13F/M44F-Az, M13F/G116F-Az, and M44F/G116F-Az combined with that of G116F-Az reveal these changes arise from steric effects and fine-tuning of hydrogen bond networks around the copper-binding His117 residue. The insights gained from this study would provide another step toward the development of redox-active proteins with tunable redox properties for many biological and biotechnological applications.

Diverse Catalytic Reactions for the Stereoselective Synthesis of Cyclic Dinucleotide MK-1454.

As practitioners of organic chemistry strive to deliver efficient syntheses of the most complex natural products and drug candidates, further innovations in synthetic strategies are required to facilitate their efficient construction. These aspirational breakthroughs often go hand-in-hand with considerable reductions in cost and environmental impact. Enzyme-catalyzed reactions have become an impressive and necessary tool that offers benefits such as increased selectivity and waste limitation. These benefits are amplified when enzymatic processes are conducted in a cascade in combination with novel bond-forming strategies. In this article, we report a highly diastereoselective synthesis of MK-1454, a potent agonist of the stimulator of interferon gene (STING) signaling pathway. The synthesis begins with the asymmetric construction of two fluoride-bearing deoxynucleotides. The routes were designed for maximum convergency and selectivity, relying on the same benign electrophilic fluorinating reagent. From these complex subunits, four enzymes are used to construct the two bridging thiophosphates in a highly selective, high yielding cascade process. Critical to the success of this reaction was a thorough understanding of the role transition metals play in bond formation.

Deep Dive into Machine Learning Models for Protein Engineering.

Protein redesign and engineering has become an important task in pharmaceutical research and development. Recent advances in technology have enabled efficient protein redesign by mimicking natural evolutionary mutation, selection, and amplification steps in the laboratory environment. For any given protein, the number of possible mutations is astronomical. It is impractical to synthesize all sequences or even to investigate all functionally interesting variants. Recently, there has been an increased interest in using machine learning to assist protein redesign, since prediction models can be used to virtually screen a large number of novel sequences. However, many state-of-the-art machine learning models, especially deep learning models, have not been extensively explored. Moreover, only a small selection of protein sequence descriptors has been considered. In this work, the performance of prediction models built using an array of machine learning methods and protein descriptor types, including two novel, single amino acid descriptors and one structure-based three-dimensional descriptor, is benchmarked. The predictions were evaluated on a diverse collection of public and proprietary data sets, using a variety of evaluation metrics. The results of this comparison suggest that Convolution Neural Network models built with amino acid property descriptors are the most widely applicable to the types of protein redesign problems faced in the pharmaceutical industry.

Design of a single protein that spans the entire 2-V range of physiological redox potentials.

The reduction potential (E°') is a critical parameter in determining the efficiency of most biological and chemical reactions. Biology employs three classes of metalloproteins to cover the majority of the 2-V range of physiological E°'s. An ultimate test of our understanding of E°' is to find out the minimal number of proteins and their variants that can cover this entire range and the structural features responsible for the extreme E°'. We report herein the design of the protein azurin to cover a range from +970 mV to -954 mV vs. standard hydrogen electrode (SHE) by mutating only five residues and using two metal ions. Spectroscopic methods have revealed geometric parameters important for the high E°'. The knowledge gained and the resulting water-soluble redox agents with predictable E°'s, in the same scaffold with the same surface properties, will find wide applications in chemical, biochemical, biophysical, and biotechnological fields.

A Purple Cupredoxin from Nitrosopumilus maritimus Containing a Mononuclear Type 1 Copper Center with an Open Binding Site.

Mononuclear cupredoxin proteins usually contain a coordinately saturated type 1 copper (T1Cu) center and function exclusively as electron carriers. Here we report a cupredoxin isolated from the nitrifying archaeon Nitrosopumilus maritimus SCM1, called Nmar1307, that contains a T1Cu center with an open binding site containing water. It displays a deep purple color due to strong absorptions around 413 nm (1880 M(-1) cm(-1)) and 558 nm (2290 M(-1) cm(-1)) in the UV-vis electronic spectrum. EPR studies suggest the protein contains two Cu(II) species of nearly equal population, one nearly axial, with hyperfine constant A∥ = 98 × 10(-4) cm(-1), and another more rhombic, with a smaller A∥ value of 69 × 10(-4) cm(-1). The X-ray crystal structure at 1.6 Å resolution confirms that it contains a Cu atom coordinated by two His and one Cys in a trigonal plane, with an axial H2O at 2.25 Å. Both UV-vis absorption and EPR spectroscopic studies suggest that the Nmar1307 can oxidize NO to nitrite, an activity that is attributable to the high reduction potential (354 mV vs SHE) of the copper site. These results suggest that mononuclear cupredoxins can have a wide range of structural features, including an open binding site containing water, making this class of proteins even more versatile.

Designed azurins show lower reorganization free energies for intraprotein electron transfer.

Low reorganization free energies are necessary for fast electron transfer (ET) reactions. Hence, rational design of redox proteins with lower reorganization free energies has been a long-standing challenge, promising to yield a deeper understanding of the underlying principles of ET reactivity and to enable potential applications in different energy conversion systems. Herein we report studies of the intramolecular ET from pulse radiolytically produced disulfide radicals to Cu(II) in rationally designed azurin mutants. In these mutants, the copper coordination sphere has been fine-tuned to span a wide range of reduction potentials while leaving the metal binding site effectively undisrupted. We find that the reorganization free energies of ET within the mutants are indeed lower than that of WT azurin, increasing the intramolecular ET rate constants almost 10-fold: changes that are correlated with increased flexibility of their copper sites. Moreover, the lower reorganization free energy results in the ET rate constants reaching a maximum value at higher driving forces, as predicted by the Marcus theory.

Axial interactions in the mixed-valent CuA active site and role of the axial methionine in electron transfer.

Within Cu-containing electron transfer active sites, the role of the axial ligand in type 1 sites is well defined, yet its role in the binuclear mixed-valent CuA sites is less clear. Recently, the mutation of the axial Met to Leu in a CuA site engineered into azurin (CuA Az) was found to have a limited effect on E(0) relative to this mutation in blue copper (BC). Detailed low-temperature absorption and magnetic circular dichroism, resonance Raman, and electron paramagnetic resonance studies on CuA Az (WT) and its M123X (X = Q, L, H) axial ligand variants indicated stronger axial ligation in M123L/H. Spectroscopically validated density functional theory calculations show that the smaller ΔE(0) is attributed to H2O coordination to the Cu center in the M123L mutant in CuA but not in the equivalent BC variant. The comparable stabilization energy of the oxidized over the reduced state in CuA and BC (CuA ∼ 180 mV; BC ∼ 250 mV) indicates that the S(Met) influences E(0) similarly in both. Electron delocalization over two Cu centers in CuA was found to minimize the Jahn-Teller distortion induced by the axial Met ligand and lower the inner-sphere reorganization energy. The Cu-S(Met) bond in oxidized CuA is weak (5.2 kcal/mol) but energetically similar to that of BC, which demonstrates that the protein matrix also serves an entatic role in keeping the Met bound to the active site to tune down E(0) while maintaining a low reorganization energy required for rapid electron transfer under physiological conditions.

Design of functional metalloproteins.

Metalloproteins catalyse some of the most complex and important processes in nature, such as photosynthesis and water oxidation. An ultimate test of our knowledge of how metalloproteins work is to design new metalloproteins. Doing so not only can reveal hidden structural features that may be missing from studies of native metalloproteins and their variants, but also can result in new metalloenzymes for biotechnological and pharmaceutical applications. Although it is much more challenging to design metalloproteins than non-metalloproteins, much progress has been made in this area, particularly in functional design, owing to recent advances in areas such as computational and structural biology.

Rationally tuning the reduction potential of a single cupredoxin beyond the natural range.

Redox processes are at the heart of numerous functions in chemistry and biology, from long-range electron transfer in photosynthesis and respiration to catalysis in industrial and fuel cell research. These functions are accomplished in nature by only a limited number of redox-active agents. A long-standing issue in these fields is how redox potentials are fine-tuned over a broad range with little change to the redox-active site or electron-transfer properties. Resolving this issue will not only advance our fundamental understanding of the roles of long-range, non-covalent interactions in redox processes, but also allow for design of redox-active proteins having tailor-made redox potentials for applications such as artificial photosynthetic centres or fuel cell catalysts for energy conversion. Here we show that two important secondary coordination sphere interactions, hydrophobicity and hydrogen-bonding, are capable of tuning the reduction potential of the cupredoxin azurin over a 700 mV range, surpassing the highest and lowest reduction potentials reported for any mononuclear cupredoxin, without perturbing the metal binding site beyond what is typical for the cupredoxin family of proteins. We also demonstrate that the effects of individual structural features are additive and that redox potential tuning of azurin is now predictable across the full range of cupredoxin potentials.

Systematic tuning of heme redox potentials and its effects on O2 reduction rates in a designed oxidase in myoglobin.

Cytochrome c Oxidase (CcO) is known to catalyze the reduction of O2 to H2O efficiently with a much lower overpotential than most other O2 reduction catalysts. However, methods by which the enzyme fine-tunes the reduction potential (E°) of its active site and the corresponding influence on the O2 reduction activity are not well understood. In this work, we report systematic tuning of the heme E° in a functional model of CcO in myoglobin containing three histidines and one tyrosine in the distal pocket of heme. By removing hydrogen-bonding interactions between Ser92 and the proximal His ligand and a heme propionate, and increasing hydrophobicity of the heme pocket through Ser92Ala mutation, we have increased the heme E° from 95 ± 2 to 123 ± 3 mV. Additionally, replacing the native heme b in the CcO mimic with heme a analogs, diacetyl, monoformyl, and diformyl hemes, that posses electron-withdrawing groups, resulted in higher E° values of 175 ± 5, 210 ± 6, and 320 ± 10 mV, respectively. Furthermore, O2 consumption studies on these CcO mimics revealed a strong enhancement in O2 reduction rates with increasing heme E°. Such methods of tuning the heme E° through a combination of secondary sphere mutations and heme substitutions can be applied to tune E° of other heme proteins, allowing for comprehensive investigations of the relationship between E° and enzymatic activity.

Spectroscopic and DFT studies of second-sphere variants of the type 1 copper site in azurin: covalent and nonlocal electrostatic contributions to reduction potentials.

The reduction potentials (E(0)) of type 1 (T1) or blue copper (BC) sites in proteins and enzymes with identical first coordination spheres around the redox active copper ion can vary by ~400 mV. Here, we use a combination of low-temperature electronic absorption and magnetic circular dichroism, electron paramagnetic resonance, resonance Raman, and S K-edge X-ray absorption spectroscopies to investigate a series of second-sphere variants--F114P, N47S, and F114N in Pseudomonas aeruginosa azurin--which modulate hydrogen bonding to and protein-derived dipoles nearby the Cu-S(Cys) bond. Density functional theory calculations correlated to the experimental data allow for the fractionation of the contributions to tuning E(0) into covalent and nonlocal electrostatic components. These are found to be significant, comparable in magnitude, and additive for active H-bonds, while passive H-bonds are mostly nonlocal electrostatic in nature. For dipoles, these terms can be additive to or oppose one another. This study provides a methodology for uncoupling covalency from nonlocal electrostatics, which, when coupled to X-ray crystallographic data, distinguishes specific local interactions from more long-range protein/active interactions, while affording further insight into the second-sphere mechanisms available to the protein to tune the E(0) of electron-transfer sites in biology.

Tuning reduction potentials of type 1 copper center in azurin by replacing a histidine ligand with its isostructural analogues.

Type 1 copper proteins have a conserved ligand set of one cysteine and two histidines, with many proteins, such as azurin, also containing an axial methionine. While the cysteine and methionine in azurin have been replaced with their respective isostructural analogues of unnatural amino acids to reveal their roles in tuning electronic structures and functional properties, such as reduction potentials (E°'), the histidine ligands have not been probed in this way. We herein report the substitution of His117 in azurin with three unnatural isostructural analogues, 5-nitrohistidine(Ntr), thiazolylalanine(SHis) and 1-methylhistidine(MeH) by expressed protein ligation. While UV-vis absorption and electron paramagnetic resonance spectroscopies confirm that isostructural replacement results in minimal structural change in the Cu(II) state, the E°' of these variants increases with increasing pKa of the δ nitrogens of the imidazole. This counter-intuitive relationship between E°' of the protein and pKa of the sidechain group suggests additional factors may play a role in tuning E°'.

Kinetics of copper incorporation into a biosynthetic purple Cu(A) azurin: characterization of red, blue, and a new intermediate species.

Evolutionary links between type 1 blue copper (T1 Cu), type 2 red copper (T2 Cu), and purple Cu(A) cupredoxins have been proposed, but the structural features and mechanism responsible for such links as well as for assembly of Cu(A) sites in vivo are poorly understood, even though recent evidence demonstrated that the Cu(II) oxidation state plays an important role in this process. In this study, we examined the kinetics of Cu(II) incorporation into the Cu(A) site of a biosynthetic Cu(A) model, Cu(A) azurin (Cu(A)Az) and found that both T1 Cu and T2 Cu intermediates form on the path to final Cu(A) reconstitution in a pH-dependent manner, with slower kinetics and greater accumulation of the intermediates as the pH is raised from 5.0 to 7.0. While these results are similar to those observed previously in the native Cu(A) center of nitrous oxide reductase, the faster kinetics of copper incorporation into Cu(A)Az allowed us to use lower copper equivalents to reveal a new pathway of copper incorporation, including a novel intermediate that has not been reported in cupredoxins before, with intense electronic absorption maxima at ~410 and 760 nm. We discovered that this new intermediate underwent reduction to Cu(I), and proposed that it is a Cu(II)-dithiolate species. Oxygen-dependence studies demonstrated that the T1 Cu species only formed in the presence of molecular oxygen, suggesting the T1 Cu intermediate is a one-electron oxidation product of a Cu(I) species. By studying Cu(A)Az variants where the Cys and His ligands are mutated, we have identified the T2 Cu intermediate as a capture complex with Cys116 and the T1 Cu intermediate as a complex with Cys112 and His120. These results led to a unified mechanism of copper incorporation and new insights regarding the evolutionary link between all cupredoxin sites as well as the in vivo assembly of Cu(A) centers.

Mutation Maker, An Open Source Oligo Design Platform for Protein Engineering.

Protein engineering is the discipline of developing useful proteins for applications in research, therapeutic, and industrial processes by modification of naturally occurring proteins or by invention of de novo proteins. Modern protein engineering relies on the ability to rapidly generate and screen diverse libraries of mutant proteins. However, design of mutant libraries is typically hampered by scale and complexity, necessitating development of advanced automation and optimization tools that can improve efficiency and accuracy. At present, automated library design tools are functionally limited or not freely available. To address these issues, we developed Mutation Maker, an open source mutagenic oligo design software for large-scale protein engineering experiments. Mutation Maker is not only specifically tailored to multisite random and directed mutagenesis protocols, but also pioneers bespoke mutagenic oligo design for de novo gene synthesis workflows. Enabled by a novel bundle of orchestrated heuristics, optimization, constraint-satisfaction and backtracking algorithms, Mutation Maker offers a versatile toolbox for gene diversification design at industrial scale. Supported by in silico simulations and compelling experimental validation data, Mutation Maker oligos produce diverse gene libraries at high success rates irrespective of genes or vectors used. Finally, Mutation Maker was created as an extensible platform on the notion that directed evolution techniques will continue to evolve and revolutionize current and future-oriented applications.

Transforming a blue copper into a red copper protein: engineering cysteine and homocysteine into the axial position of azurin using site-directed mutagenesis and expressed protein ligation.

Interactions of the axial ligand with its blue copper center are known to be important in tuning spectroscopic and redox properties of cupredoxins. While conversion of the blue copper center with a weak axial ligand to a green copper center containing a medium strength axial ligand has been demonstrated in cupredoxins, converting the blue copper center to a red copper center with a strong axial ligand has not been reported. Here we show that replacing Met121 in azurin from Pseudomonas aeruginosa with Cys caused an increased ratio (R(L)) of absorption at 447 nm over that at 621 nm. Whereas no axial Cu-S(Cys121) interaction in Met121Cys was detectable by extended X-ray absorption fine structure (EXAFS) spectroscopy at pH 5, similar to what was observed in native azurin with Met121 as the axial ligand, the Cu-S(Cys121) interaction at 2.74 A is clearly visible at higher pH. Despite the higher R(L) and stronger axial Cys121 interaction with Cu(II) ion, the Met121Cys variant remains largely a type 1 copper protein at low pH (with hyperfine coupling constant A( parallel) = 54 x 10(-4) cm(-1) at pH 4 and 5), or distorted type 1 or green copper protein at high pH (A(parallel) = 87 x 10(-4) cm(-1) at pH 8 and 9), attributable to the relatively long distance between the axial ligand and copper and the constraint placed by the protein scaffold. To shorten the distance between axial ligand and copper, we replaced Met121 with a nonproteinogenic amino acid homocysteine that contains an extra methylene group, resulting in a variant whose spectra (R(L)= 1.5, and A(parallel) = 180 x 10(-4) cm(-1)) and Cu-S(Cys) distance (2.22 A) are very similar to those of the red copper protein nitrosocyanin. Replacing Met121 with Cys or homocysteine resulted in lowering of the reduction potential from 222 mV in the native azurin to 95 +/- 3 mV for Met121Cys azurin and 113 +/- 6 mV for Met121Hcy azurin at pH 7. The results strongly support the "coupled distortion" model that helps explain axial ligand tuning of spectroscopic properties in cupredoxins, and demonstrate the power of using unnatural amino acids to address critical chemical biological questions.

Computer-based Engineering of Thermostabilized Antibody Fragments.

We used the molecular modeling program Rosetta to identify clusters of amino acid substitutions in antibody fragments (scFvs and scAbs) that improve global protein stability and resistance to thermal deactivation. Using this methodology, we increased the melting temperature (Tm) and resistance to heat treatment of an antibody fragment that binds to the Clostridium botulinum hemagglutinin protein (anti-HA33). Two designed antibody fragment variants with two amino acid replacement clusters, designed to stabilize local regions, were shown to have both higher Tm compared to the parental scFv and importantly, to retain full antigen binding activity after 2 hours of incubation at 70 °C. The crystal structure of one thermostabilized scFv variants was solved at 1.6 Å and shown to be in close agreement with the RosettaAntibody model prediction.

Design of an in vitro biocatalytic cascade for the manufacture of islatravir.

Enzyme-catalyzed reactions have begun to transform pharmaceutical manufacturing, offering levels of selectivity and tunability that can dramatically improve chemical synthesis. Combining enzymatic reactions into multistep biocatalytic cascades brings additional benefits. Cascades avoid the waste generated by purification of intermediates. They also allow reactions to be linked together to overcome an unfavorable equilibrium or avoid the accumulation of unstable or inhibitory intermediates. We report an in vitro biocatalytic cascade synthesis of the investigational HIV treatment islatravir. Five enzymes were engineered through directed evolution to act on non-natural substrates. These were combined with four auxiliary enzymes to construct islatravir from simple building blocks in a three-step biocatalytic cascade. The overall synthesis requires fewer than half the number of steps of the previously reported routes.

Reversible S-nitrosylation in an engineered azurin.

S-Nitrosothiols are known as reagents for NO storage and transportation and as regulators in many physiological processes. Although the S-nitrosylation catalysed by haem proteins is well known, no direct evidence of S-nitrosylation in copper proteins has been reported. Here, we report reversible insertion of NO into a copper-thiolate bond in an engineered copper centre in Pseudomonas aeruginosa azurin by rational design of the primary coordination sphere and tuning its reduction potential by deleting a hydrogen bond in the secondary coordination sphere. The results not only provide the first direct evidence of S-nitrosylation of Cu(II)-bound cysteine in metalloproteins, but also shed light on the reaction mechanism and structural features responsible for stabilizing the elusive Cu(I)-S(Cys)NO species. The fast, efficient and reversible S-nitrosylation reaction is used to demonstrate its ability to prevent NO inhibition of cytochrome bo3 oxidase activity by competing for NO binding with the native enzyme under physiologically relevant conditions.

An alternate pathway of arsenate resistance in E. coli mediated by the glutathione S-transferase GstB.

Microbial arsenate resistance is known to be conferred by specialized oxidoreductase enzymes termed arsenate reductases. We carried out a genetic selection on media supplemented with sodium arsenate for multicopy genes that can confer growth to E. coli mutant cells lacking the gene for arsenate reductase (E. coli ΔarsC). We found that overexpression of glutathione S-transferase B (GstB) complemented the ΔarsC allele and conferred growth on media containing up to 5 mM sodium arsenate. Interestingly, unlike wild type E. coli arsenate reductase, arsenate resistance via GstB was not dependent on reducing equivalents provided by glutaredoxins or a catalytic cysteine residue. Instead, two arginine residues, which presumably coordinate the arsenate substrate within the electrophilic binding site of GstB, were found to be critical for transferase activity. We provide biochemical evidence that GstB acts to directly reduce arsenate to arsenite with reduced glutathione (GSH) as the electron donor. Our results reveal a pathway for the detoxification of arsenate in bacteria that hinges on a previously undescribed function of a bacterial glutathione S-transferase.