Deep learning framework for peak detection at the intact level of therapeutic proteins.

While automated peak detection functionalities are available in commercially accessible software, achieving optimal true positive rates frequently necessitates visual inspection and manual adjustments. In the initial phase of this study, hetero-variants (glycoforms) of a monoclonal antibody were distinguished using liquid chromatography-mass spectrometry, revealing discernible peaks at the intact level. To comprehensively identify each peak (hetero-variant) in the intact-level analysis, a deep learning approach utilizing convolutional neural networks (CNNs) was employed in the subsequent phase of the study. In the current case study, utilizing conventional software for peak identification, five peaks were detected using a 0.5 threshold, whereas seven peaks were identified using the CNN model. The model exhibited strong performance with a probability area under the curve (AUC) of 0.9949, surpassing that of partial least squares discriminant analysis (PLS-DA) (probability AUC of 0.8041), and locally weighted regression (LWR) (probability AUC of 0.6885) on the data acquired during experimentation in real-time. The AUC of the receiver operating characteristic curve also illustrated the superior performance of the CNN over PLS-DA and LWR.

Improved Stability and Manufacturability of Nucleocapsid Antigens for SARS-CoV2 Diagnostics through Protein Engineering.

The COVID-19 pandemic has had a significant impact on human health management. A rapid diagnosis of SARS-CoV2 at the point-of-care (POC) is critical to prevent disease spread. As a POC device for remote settings, a LFIA should not require cold-chain maintenance and should be kept at normal temperatures. Antigen stability can be enhanced by addressing instability issues when dealing with fragile components, such as proteinaceous capture antigens. This study used immunologically guided protein engineering to enhance the capture nucleocapsid (NP) antigen stability of SARS-CoV2. A search of the IEDB database revealed that antibodies detecting epitopes are almost uniformly distributed over NP1-419. In contrast, N-terminal stretches of NP1-419 are theoretically more unstable than C-terminal stretches. We identified NP250-365 as a NP stretch with a low instability index and B-cell epitopes. Apart from NP1-419, two other variants (NP121-419 and NP250-365) were cloned, expressed, and purified. The degradation pattern of the proteins was observed on SDS-PAGE after three days of stability studies at -20 °C, 4 °C, and 37 °C. NP1-419 was the most degraded while NP250-365 exhibited the least degradation. Also, NP1-419, NP250-365, and NP121-419 reacted with purified antibodies from COVID-19 patient serum. Our results suggest that NP250-365 may be used as a stable capture antigen in LFIA devices to detect COVID-19.

Assessment of structural and functional similarity of biosimilar products: Bevacizumab as a case study.

The antiangiogenic drug bevacizumab is a blockbuster therapeutic pharmaceutical product that is used to treat many different types of cancer including kidney, colon, rectum, lung, and breast cancer. As a result, multiple biosimilars have been approved across the various regulatory jurisdictions in India (>20 in number till date). The rapidly growing market and acceptance of biosimilars was the motivation to perform comparability study of bevacizumab biosimilars that are presently available in the Indian market. A comprehensive analytical and functional biosimilarity assessment has been performed to examine and compare innovator product of bevacizumab (Avastin-innovator product, Roche Products (India) Pvt Ltd) and six biosimilars that are being marketed in India (Abevmy from Mylan Pharmaceuticals Pvt Ltd, Bevazza from Lupin Ltd, Bryxta from Zydus Cadila, Krabeva from Biocon, Ivzumab from RPG Life Sciences Ltd, and Advamab from Alkem Laboratories Ltd). Physiochemical characterization of drug products was performed with respect to their primary structure (intact mass, reduced mass, peptide mapping by LC-MS), higher order structure (secondary structure by FTIR, Far-UV-CD, and tertiary structure by Near-UV-CD, intrinsic fluorescence spectroscopy), impurity profile (SE-HPLC, SEC-MALS, extrinsic fluorescence: size heterogenicity, degradation, stability; DLS: hydrodynamic radius; WCX-HPLC: charge variants analysis) and post-translational modifications by measuring reduced glycans through fluorescence dye analysis. Functional characterization was performed by SPR and cell proliferation assay. Further, chemometrics based quantitative evaluation of biosimilarity has been performed by combining the data obtained from analytical characterization platform. The analysis of the analytical, functional and chemometric results revealed significant levels of similarity, with biosimilar4 being the sole exception. Despite being within product specifications, Biosimilar4 displayed significant deviations with respect to critical quality attributes, including a lower proportion of monomer content, a larger percentage of basic charge variant species, and a lower proportion of aglycosylated glycoform.

A novel filter-assisted protein precipitation (FAPP) based sample pre-treatment method for LC-MS peptide mapping for biosimilar characterization.

Establishing analytical and functional comparability serves as the foundation of biosimilar development. A critical part of this exercise is sequence similarity search and categorization of post-translational modifications (PTMs), often by peptide mapping using liquid chromatography-mass spectrometry (LC-MS). When performing bottom-up proteomic sample preparation, efficient digestion of the protein and extraction of peptides for subsequent mass spectrometric analysis can be a challenge. Conventional sample preparation strategies face the risk of allowing interference of chemicals which are essential for extraction but are likely to interfere with digestion, resulting in complex chromatographic profiles due to semi-cleavages, insufficient peptide cleavages, and other unwanted reactions. Further, peptide cleanup through commonly used immobilized C-18 pipette tips can cause significant peptide loss as well as variability in individual peptide yields, thereby causing artifacts of various product-related modifications. In this study, we proposed a simple enzymatic digestion technique by incorporating different molecular weight filters and protein precipitation, with the objective to minimize interference of denaturing, reducing, and alkylating agents throughout overnight digestion. As a result, the need for peptide cleanup is significantly reduced and results in higher peptide yield. The proposed FAPP approach outperformed the conventional method across multiple metrics including, 30% more peptides, 8.19% more fully digested peptides, 14% higher sequence coverage rate, and 11.82% more site-specific alterations. Quantitative and qualitative repeatability of the proposed approach have been demonstrated. It can be concluded that the filter-assisted protein precipitation (FAPP) protocol proposed in this study offers an effective substitute for the traditional approach.

A native multi-dimensional monitoring workflow for at-line characterization of mAb titer, size, charge, and glycoform heterogeneities in cell culture supernatant.

With growing maturity of the biopharmaceutical industry, new modalities entering the therapeutic design space and increasing complexity of formulations such as combination therapy, the demands and requirements on analytical workflows have also increased. A recent evolution in newer analytical workflows is that of multi-attribute monitoring workflows designed on chromatography-mass spectrometry (LC-MS) platform. In comparison to traditional one attribute per workflow paradigm, multi-attribute workflows are designed to monitor multiple critical quality attributes through a single workflow, thus reducing the overall time to information and increasing efficiency and throughput. While the 1st generation multi-attribute workflows focused on bottom-up characterization following peptide digestion, the more recent workflows have been focussing on characterization of intact biologics, preferably in native state. So far intact multi-attribute monitoring workflows suitable for comparability, utilizing single dimension chromatography coupled with MS have been published. In this study, we describe a native multi-dimensional multi-attribute monitoring workflow for at-line characterization of monoclonal antibody (mAb) titer, size, charge, and glycoform heterogeneities directly in cell culture supernatant. This has been achieved through coupling ProA in series with size exclusion chromatography in 1st dimension followed by cation exchange chromatography in the 2nd dimension. Intact paired glycoform characterization has been achieved through coupling 2D-LC with q-ToF-MS. The workflow with a single heart cut can be completed in 25 mins and utilizes 2D-liquid chromatography (2D-LC) to maximize separation and monitoring of titer, size as well as charge variants.

Taking the individual bias out of examining comparability of biosimilars: A case study on monoclonal antibody therapeutics.

Biosimilar manufacturers need to perform analytical and functional similarity assessments against the reference product. Successful demonstration allows for an abbreviated clinical path, thereby translating to affordable biosimilars. Current practices for regulatory concurrence on analytical similarity data are based on chart visualization and open to individual (human) bias. Here, we present a novel, chemometric approach for assessing biosimilarity that aims to simplify assessment and eliminate individual bias from decision making through application of weighted principal component analysis. Through the proposed approach, chemical information across the analytical characterization platform and drug products can be collated into a single plot for quantitative biosimilarity assessment. The proposed one-plot analysis offers a holistic visualization of 1) inter-product variability (w.r.t reference product) in cases where multiple batches per product have been investigated and 2) intra-product variability for each critical quality attribute (CQA) wherein information from orthogonal tools can be incorporated within the same plot. This allows for numerical grading of similarity for biosimilars of any given reference product. Although the proposed statistical approach is novel, it builds on standardized measures of CQA, criticality, and analytical procedures, thus making this approach easy to incorporate within the existing regulatory framework.

Characterization of haemoglobin from actinorhizal plants--an in silico approach.

Plant haemoglobins (Hbs), found in both symbiotic and non-symbiotic plants, are heme proteins and members of the globin superfamily. Hb genes of actinorhizal Fagales mostly belong to the non-symbiotic type of haemoglobin; however, along with the non-symbiotic Hb, Casuarina sp. posses a symbiotic one (symCgHb), which is expressed specifically in infected cells of nodules. A thorough sequence analysis of 26 plant Hb proteins, currently available in public domain, revealed a consensus motif of 29 amino acids. This motif is present in all the members of symbiotic class II Hbs including symCgHb and non-symbiotic Class II Hbs, but is totally absent in Class I symbiotic and non-symbiotic Hbs. Further, we constructed 3D structures of Hb proteins from Alnus and Casuarina through homology modelling and peeped into their structural properties. Structure-based studies revealed that the Casuarina symbiotic haemoglobin protein shows distinct stereochemical properties from that of the other Casuarina and Alnus Hb proteins. It also showed considerable structural similarities with leghemoglobin structure from yellow lupin (pdb id 1GDI). Therefore, sequence and structure analyses point to the fact that symCgHb protein shows significant resemblance to symbiotic haemoglobin found in legumes and may thus eventually play a similar role in shielding the nitrogenase from oxygen as seen in the case of leghemoglobin.

A database for Mycobacterium secretome analysis: 'MycoSec' to accelerate global health research.

Abstract Members of the genus Mycobacterium are notorious for their pathogenesis. Investigations from various perspectives have identified the pathogenic strategies employed by these lethal pathogens. Secretomes are believed to play crucial roles in host cell recognition and cross-talks, in cellular attachment, and in triggering other functions related to host pathogen interactions. However, a proper idea of the mycobacterial secretomes and their mechanism of functionality still remains elusive. In the present study, we have developed a comprehensive database of potential mycobacterial secretomes (MycoSec) using pre-existing algorithms for secretome prediction for researchers interested in this particular field. The database provides a platform for retrieval and analysis of identified secretomes in all finished genomes of the family Mycobacteriaceae. The database contains valuable information regarding secretory signal peptides (Sec type), lipoprotein signal peptides (Lipo type), and Twin arginine (RR/KR) signal peptides (TAT type), prevalent in mycobacteria. Information pertaining to COG analysis, codon usage, and gene expression of the predicted secretomes has also been incorporated in the database. MycoSec promises to be a useful repertoire providing a plethora of information regarding mycobacterial secretomes and may well be a platform to speed global health research. MycoSec is freely accessible at http://www.bicnbu.in/mycosec .