Deep learning image analysis models pretrained on daily objects are useful for the preliminary characterization of particulate pharmaceutical samples.

Visible and subvisible particles are a quality attribute in sterile pharmaceutical samples. A common method for characterizing and quantifying pharmaceutical samples containing particulates is imaging many individual particles using high-throughput instrumentation and analyzing the populations data. The analysis includes conventional metrics such as the particle size distribution but can be more sophisticated by interpreting other visual/morphological features. To avoid the hurdles of building new image analysis models capable of extracting such relevant features from scratch, we propose using well-established pretrained deep learning image analysis models such as EfficientNet. We demonstrate that such models are useful as a prescreening tool for high-level characterization of biopharmaceutical particle image data. We show that although these models are originally trained for completely different tasks (such as the classification of daily objects in the ImageNet database), the visual feature vectors extracted by such models can be useful for studying different types of subvisible particles. This applicability is illustrated through multiple case studies: (i) particle risk assessment in prefilled syringe formulations containing different particle types such as silicone oil, (ii) method comparability with the example of accelerated forced degradation, and (iii) excipient influence on particle morphology with the example of Polysorbate 80 (PS80). As examples of agnostic applicability of pretrained models, we also elucidate the application to two high-throughput microscopy methods: microflow and background membrane imaging. We show that different particle populations with different morphological and visual features can be identified in different samples by leveraging out-of-the-box pretrained models to analyze images from each sample.

Characterizing Silicone Oil-Induced Protein Aggregation with Stimulated Raman Scattering Imaging.

Particles in biopharmaceutical products present high risks due to their detrimental impacts on product quality and safety. Identification and quantification of particles in drug products are important to understand particle formation mechanisms, which can help develop control strategies for particle formation during the formulation development and manufacturing process. However, existing analytical techniques such as microflow imaging and light obscuration measurement lack the sensitivity and resolution to detect particles with sizes smaller than 2 µm. More importantly, these techniques are not able to provide chemical information to determine particle composition. In this work, we overcome these challenges by applying the stimulated Raman scattering (SRS) microscopy technique to monitor the C-H Raman stretching modes of the proteinaceous particles and silicone oil droplets formed in the prefilled syringe barrel. By comparing the relative signal intensity and spectral features of each component, most particles can be classified as protein-silicone oil aggregates. We further show that morphological features are poor indicators of particle composition. Our method has the capability to quantify aggregation in protein therapeutics with chemical and spatial information in a label-free manner, potentially allowing high throughput screening or investigation of aggregation mechanisms.

Searching for the low affinity ubiquinone binding site in cytochrome bo3 from Escherichia coli.

The cytochrome bo3 ubiquinol oxidase is one of three respiratory oxygen reductases in the aerobic respiratory chain of Escherichia coli. The generally accepted model of catalysis assumes that cyt bo3 contains two distinct ubiquinol binding sites: (i) a low affinity (QL) site which is the traditional substrate binding site; and (ii) a high affinity (QH) site where a "permanently" bound quinone acts as a cofactor, taking two electrons from the substrate quinol and passing them one-by-one to the heme b component of the enzyme which, in turn, transfers them to the heme o3/CuB active site. Whereas the residues at the QH site are well defined, the location of the QL site remains unknown. The published X-ray structure does not contain quinone, and substantial amounts of the protein are missing as well. A recent bioinformatics study by Bossis et al. [Biochem J. (2014) 461, 305-314] identified a sequence motif G163EFX3GWX2Y173 as the likely QL site in the family of related quinol oxidases. In the current work, this was tested by site-directed mutagenesis. The results show that these residues are not important for catalytic function and do not define the QL substrate binding site.

Significant improvement of oxidase activity through the genetic incorporation of a redox-active unnatural amino acid.

While nature employs various covalent and non-covalent strategies to modulate tyrosine (Y) redox potential and pK a in order to optimize enzyme activities, such approaches have not been systematically applied for the design of functional metalloproteins. Through the genetic incorporation of 3-methoxytyrosine (OMeY) into myoglobin, we replicated important features of cytochrome c oxidase (CcO) in this small soluble protein, which exhibits selective O2 reduction activity while generating a small amount of reactive oxygen species (ROS). These results demonstrate that the electron donating ability of a tyrosine residue in the active site is important for CcO function. Moreover, we elucidated the structural basis for the genetic incorporation of OMeY into proteins by solving the X-ray structure of OMeY specific aminoacyl-tRNA synthetase complexed with OMeY.

Conformational coupling between the active site and residues within the K(C)-channel of the Vibrio cholerae cbb3-type (C-family) oxygen reductase.

The respiratory chains of nearly all aerobic organisms are terminated by proton-pumping heme-copper oxygen reductases (HCOs). Previous studies have established that C-family HCOs contain a single channel for uptake from the bacterial cytoplasm of all chemical and pumped protons, and that the entrance of the K(C)-channel is a conserved glutamate in subunit III. However, the majority of the K(C)-channel is within subunit I, and the pathway from this conserved glutamate to subunit I is not evident. In the present study, molecular dynamics simulations were used to characterize a chain of water molecules leading from the cytoplasmic solution, passing the conserved glutamate in subunit III and extending into subunit I. Formation of the water chain, which controls the delivery of protons to the K(C)-channel, was found to depend on the conformation of Y241(Vc), located in subunit I at the interface with subunit III. Mutations of Y241(Vc) (to A/F/H/S) in the Vibrio cholerae cbb3 eliminate catalytic activity, but also cause perturbations that propagate over a 28-Å distance to the active site heme b3. The data suggest a linkage between residues lining the K(C)-channel and the active site of the enzyme, possibly mediated by transmembrane helix α7, which contains both Y241(Vc) and the active site cross-linked Y255(Vc), as well as two CuB histidine ligands. Other mutations of residues within or near helix α7 also perturb the active site, indicating that this helix is involved in modulation of the active site of the enzyme.

Functional importance of a pair of conserved glutamic acid residues and of Ca(2+) binding in the cbb(3)-type oxygen reductases from Rhodobacter sphaeroides and Vibrio cholerae.

The cbb(3)-type cytochrome c oxidases are members of the family of heme-copper proton pumping respiratory oxygen reductases. The structure of the cbb(3)-type oxidase from Pseudomonas stutzeri reveals that, in addition to the six redox-active metal centers (two b-type hemes, three c-type hemes, and Cu(B)), the enzyme also contains at least one Ca(2+). The calcium bridges two propionate carboxyls at the interface between the low-spin heme b and the active-site heme b(3) and, in addition, is ligated to a serine in subunit CcoO and by a glutamate in subunit CcoN. The glutamate that is ligated to Ca(2+) is one of a pair of glutamic acid residues that has previously been suggested to be part of a proton exit pathway for pumped protons. In this work, mutations of these glutamates are investigated in the cbb(3)-type oxidases from Vibrio cholerae and Rhodobacter sphaeroides. Metal analysis shows that each of these wild-type enzymes contains Ca(2+). Mutations of the glutamate expected to ligate the Ca(2+) in each of these enzymes (E126 in V. cholerae and E180 in R. sphaeroides) result in a loss of activity as well as a loss of Ca(2+). Mutations of the nearby glutamate (E129 in V. cholerae and E183 in R. sphaeroides) also resulted in a loss of oxidase activity and a loss of Ca(2+). It is concluded that the Ca(2+) is essential for assembly of the fully functional enzyme and that neither of the glutamates is likely to be part of a pathway for pumped protons within the cbb(3)-type oxygen reductases. A more likely role for these glutamates is the maintenance of the structural integrity of the active conformation of the enzyme.

Adaptation of aerobic respiration to low O2 environments.

Aerobic respiration in bacteria, Archaea, and mitochondria is performed by oxygen reductase members of the heme-copper oxidoreductase superfamily. These enzymes are redox-driven proton pumps which conserve part of the free energy released from oxygen reduction to generate a proton motive force. The oxygen reductases can be divided into three main families based on evolutionary and structural analyses (A-, B- and C-families), with the B- and C-families evolving after the A-family. The A-family utilizes two proton input channels to transfer protons for pumping and chemistry, whereas the B- and C-families require only one. Generally, the B- and C-families also have higher apparent oxygen affinities than the A-family. Here we use whole cell proton pumping measurements to demonstrate differential proton pumping efficiencies between representatives of the A-, B-, and C-oxygen reductase families. The A-family has a coupling stoichiometry of 1 H(+)/e(-), whereas the B- and C-families have coupling stoichiometries of 0.5 H(+)/e(-). The differential proton pumping stoichiometries, along with differences in the structures of the proton-conducting channels, place critical constraints on models of the mechanism of proton pumping. Most significantly, it is proposed that the adaptation of aerobic respiration to low oxygen environments resulted in a concomitant reduction in energy conservation efficiency, with important physiological and ecological consequences.

The quinone-binding sites of the cytochrome bo3 ubiquinol oxidase from Escherichia coli.

Cytochrome bo(3) is the major respiratory oxidase located in the cytoplasmic membrane of Escherichia coli when grown under high oxygen tension. The enzyme catalyzes the 2-electron oxidation of ubiquinol-8 and the 4-electron reduction of dioxygen to water. When solubilized and isolated using dodecylmaltoside, the enzyme contains one equivalent of ubiquinone-8, bound at a high affinity site (Q(H)). The quinone bound at the Q(H) site can form a stable semiquinone, and the amino acid residues which hydrogen bond to the semiquinone have been identified. In the current work, it is shown that the tightly bound ubiquinone-8 at the Q(H) site is not displaced by ubiquinol-1 even during enzyme turnover. Furthermore, the presence of high affinity inhibitors, HQNO and aurachin C1-10, does not displace ubiquinone-8 from the Q(H) site. The data clearly support the existence of a second binding site for ubiquinone, the Q(L) site, which can rapidly exchange with the substrate pool. HQNO is shown to bind to a single site on the enzyme and to prevent formation of the stable ubisemiquinone, though without displacing the bound quinone. Inhibition of the steady state kinetics of the enzyme indicates that aurachin C1-10 may compete for binding with quinol at the Q(L) site while, at the same time, preventing formation of the ubisemiquinone at the Q(H) site. It is suggested that the two quinone binding sites may be adjacent to each other or partially overlap.