Convolutional Neural Networks Guided Raman Spectroscopy as a Process Analytical Technology (PAT) Tool for Monitoring and Simultaneous Prediction of Monoclonal Antibody Charge Variants.

BACKGROUND: Charge related heterogeneities of monoclonal antibody (mAb) based therapeutic products are increasingly being considered as a critical quality attribute (CQA). They are typically estimated using analytical cation exchange chromatography (CEX), which is time consuming and not suitable for real time control. Raman spectroscopy coupled with artificial intelligence (AI) tools offers an opportunity for real time monitoring and control of charge variants. OBJECTIVE: We present a process analytical technology (PAT) tool for on-line and real-time charge variant determination during process scale CEX based on Raman spectroscopy employing machine learning techniques. METHOD: Raman spectra are collected from a reference library of samples with distribution of acidic, main, and basic species from 0-100% in a mAb concentration range of 0-20 g/L generated from process-scale CEX. The performance of different machine learning techniques for spectral processing is compared for predicting different charge variant species. RESULT: A convolutional neural network (CNN) based model was successfully calibrated for quantification of acidic species, main species, basic species, and total protein concentration with R2 values of 0.94, 0.99, 0.96 and 0.99, respectively, and the Root Mean Squared Error (RMSE) of 0.1846, 0.1627, and 0.1029 g/L, respectively, and 0.2483 g/L for the total protein concentration. CONCLUSION: We demonstrate that Raman spectroscopy combined with AI-ML frameworks can deliver rapid and accurate determination of product related impurities. This approach can be used for real time CEX pooling decisions in mAb production processes, thus enabling consistent charge variant profiles to be achieved.

Continuous integrated manufacturing for biopharmaceuticals: A new paradigm or an empty promise?

Continuous integrated bioprocessing has elicited considerable interest from the biopharma industry for the many purported benefits it promises. Today many major biopharma manufacturers around the world are engaged in the development of continuous process platforms for their products. In spite of great potential, the path toward continuous integrated bioprocessing remains unclear for the biologics industry due to legacy infrastructure, process integration challenges, vague regulatory guidelines, and a diverging focus toward novel therapies. In this article, we present a review and perspective on this topic. We explore the status of the implementation of continuous integrated bioprocessing among biopharmaceutical manufacturers. We also present some of the key hurdles that manufacturers are likely to face during this implementation. Finally, we hypothesize that the real impact of continuous manufacturing is likely to come when the cost of manufacturing is a substantial portion of the cost of product development, such as in the case of biosimilar manufacturing and emerging economies.

FRET-GP - A Local Measure of the Impact of Transmembrane Peptide on Lipids.

Reconstitution of a transmembrane protein in model lipid systems allows studying its structure and dynamics in isolation from the complexity of the natural environment. This approach also provides a well-defined environment for studying the interactions of proteins with lipids. In this work, we describe the FRET-GP method, which utilizes Förster resonance energy transfer (FRET) to specifically probe the nanoenvironment of a transmembrane domain. The tryptophan residues flanking this domain act as efficient FRET donors, while Laurdan acts as acceptor. The fluorescence of this solvatochromic probe is quantified using generalized polarization (GP) to report on lipid mobility in the vicinity of the transmembrane domain. We applied FRET-GP to study the transmembrane peptide WALP incorporated in liposomes. We found that the direct excitation of Laurdan to its second singlet state strongly contributes to GP values measured in FRET conditions. Removal of this parasitic contribution was essential for proper determination of GPFRET - the local analogue of classical GP parameter. The presence of WALP significantly increased both parameters but the local effects were considerably stronger (GPFRET ≫ GP). We conclude that WALP restricts lipid movement in its vicinity, inducing lateral inhomogeneity in membrane fluidity. WALP was also found to influence lipid phase transition. Our findings demonstrated that FRET-GP simultaneously provides local and global results, thereby enhancing the depth of information obtained from the measurement. We highlight the simplicity and sensitivity of the method, but also discuss its potential and limitations in studying protein-lipid interactions.

Control of surge tanks for continuous manufacturing of monoclonal antibodies.

Surge tanks are critical but often overlooked enablers of continuous bioprocessing. They provide multiple benefits including dampening of concentration gradients and allowing process resumption efforts in case of equipment failure or unexpected deviations, which can occur during a continuous campaign of weeks or months. They are also useful in enabling steady-state operation across a continuous train by facilitating mass balance between unit operations such as chromatography which have periodic loading and elution cycles. In this paper, we propose a design of a system of surge tanks for a monoclonal antibody (mAb) production process consisting of cell culture, clarification, capture chromatography, viral inactivation, polishing chromatography, and single-pass ultrafiltration and diafiltration. A Python controller has been developed for robust control of the continuous train. The controller has four layers, namely data acquisition, process scheduling, deviation handling, and real-time execution. A set of general guidelines for surge tank placement and sizing have been proposed together with process control strategies based on the design space of the individual unit operations, failure modes analysis of the different equipment, and expected variability in the process feed streams for both fed-batch and perfusion bioreactors. The control system has been successfully demonstrated for several continuous runs of up to 36 h in duration and is able to leverage surge tanks for robust control of the continuous train in the face of product variability as well as process errors while maintaining critical quality attributes. The proposed set of strategies for surge tank control are adaptable to most continuous processing setups for mAbs, and together form the first framework that can fully realize the benefits of surge tanks in continuous bioprocessing.

An NIR-based PAT approach for real-time control of loading in Protein A chromatography in continuous manufacturing of monoclonal antibodies.

Control of column loading in Protein A chromatography is a crucial part of development of robust and flexible process platforms for continuous production of monoclonal antibody (mAb) products. In this paper, we propose a control system that uses near infrared spectroscopy (NIRS) flow cells to accomplish the above. Two applications have been demonstrated using a periodic counter-current continuous chromatography setup. The first application involves use of single NIR flow cell before the inlet of the loading column to measure the concentration of mAb in the harvested broth. Measurement was in real-time (every 3 s) and within ±0.05 mg/ml, significantly better than making UV-based concentration estimations. The second application involved use of an additional NIR flow cell at the outlet of the loading column to measure column breakthrough in real time. The concentration data was transferred to a Python-based monitoring and control algorithm layered over a Cadence BioSMB system. The program could successfully run a three-column periodic counter current method on the BioSMB whereas controlling loading to ensure optimal resin utilization in each loading cycle phase based on precharacterized dynamic binding capacity models, whereas maintaining periodic elutions. The system was tested with multiple perturbations in harvest concentration, modeled after deviations that could arise downstream of a perfusion cell culture system. The results show that the proposed control is a spectroscopy-based process analytical technology tool that facilitates real time monitoring and control of loading in process chromatography. It is adaptable to any continuous chromatography equipment and is very well suited for implementation in a continuous mAb production train.

Process analytical technology implementation for protein refolding: GCSF as a case study.

Process analytical technology is gaining interest in the biopharmaceutical industry as a means to enable consistency in processing and thereby in product quality via process control. Protein refolding is known to be significantly impacted by critical process parameters and feed material attributes including composition and pH of the solubilisation and refolding buffers. Hence, to achieve robust process control and product quality, these attributes and parameters need to be monitored. This paper presents an approach towards statistical process control and monitoring of protein refolding, from buffer preparation to refold quenching, during manufacturing of therapeutic proteins from Escherichia coli based systems. The proposed approach utilises measurements of online redox potential, temperature, and pH for development of a statistical model. The model has then been integrated with LabView to permit real-time monitoring of the refolding process. The proposed system has been demonstrated to successfully identify process deviations and thereby enable process control for manufacturing product of consistent quality.

Investigation of pH-induced protein conformation changes by nanomechanical deflection.

Broad-spectrum biosensing technologies examine sensor signals using biomarkers, such as proteins, DNA, antibodies, specific cells, and macromolecules, based on direct- or indirect-conformational changes. Here, we have investigated the pH-dependent conformational isomerization of human serum albumin (HSA) using microcantilevers as a sensing platform. Native and denatured proteins were immobilized on cantilever surfaces to understand the effect of pH on conformational changes of the protein with respect to the coupling ligand. Our results show that protonation and deprotonation of amino acid residues on proteins play a significant role in generating charge-induced cantilever deflection. Surface plasmon resonance (SPR) was employed as a complementary technique to validate the results.

A Cyber-Physical Production System for the Integrated Operation and Monitoring of a Continuous Manufacturing Train for the Production of Monoclonal Antibodies.

The continuous manufacturing of biologics offers significant advantages in terms of reducing manufacturing costs and increasing capacity, but it is not yet widely implemented by the industry due to major challenges in the automation, scheduling, process monitoring, continued process verification, and real-time control of multiple interconnected processing steps, which must be tightly controlled to produce a safe and efficacious product. The process produces a large amount of data from different sensors, analytical instruments, and offline analyses, requiring organization, storage, and analyses for process monitoring and control without compromising accuracy. We present a case study of a cyber-physical production system (CPPS) for the continuous manufacturing of mAbs that provides an automation infrastructure for data collection and storage in a data historian, along with data management tools that enable real-time analysis of the ongoing process using multivariate algorithms. The CPPS also facilitates process control and provides support in handling deviations at the process level by allowing the continuous train to re-adjust itself via a series of interconnected surge tanks and by recommending corrective actions to the operator. Successful steady-state operation is demonstrated for 55 h with end-to-end process automation and data collection via a range of in-line and at-line sensors. Following this, a series of deviations in the downstream unit operations, including affinity capture chromatography, cation exchange chromatography, and ultrafiltration, are monitored and tracked using multivariate approaches and in-process controls. The system is in line with Industry 4.0 and smart manufacturing concepts and is the first end-to-end CPPS for the continuous manufacturing of mAbs.

Genome-wide identification of microsatellites for mapping, genetic diversity and cross-transferability in wheat (Triticum spp).

Wheat (Triticum aestivum L.) is a crucial global staple crop, and is consistently being improved to enhance yield, disease resistance, and quality traits. However, the development of molecular markers is a challenging task due to its hexaploid genome. Molecular marker system such as simple sequence repeat (SSR) and single nucleotide polymorphism (SNP) are helpful for breeding, but SNP has limitations due to its development cost and its conversion to breeder markers. The study proposed an in-silico approach, by utilizing the low-cost transcriptome sequencing of two parental lines, 'TAC 75' and 'WH 1105', to identify polymorphic SSRs for mapping in a recombinant inbred line (RIL) population. This study introduces a new approach to bridge wheat genetics intricacies and next-generation sequencing potential. It presents a comprehensive genome-wide SSR distribution using IWGSC CS RefSeq v2.1 genome assembly and to identify 189 polymorphic loci through in-silico strategy. Of these, 54.76% showed polymorphism between parents, surpassing the traditional low polymorphic success rate. A RIL population screening validated these markers, demonstrating the fitness of identified markers through chi-square tests. The designed SSRs were also validated for genetic diversity analysis in a subset of 37 Indian wheat genotypes and cross-transferability in the wild/relative wheat species. In diversity analysis, a subset of 38 markers revealed 95 alleles (2.5 allele/locus), indicating substantial genetic variation. Population structure analysis unveiled three distinct groups, supported by phylogenetic and PCoA analyses. Further the polymorphic SSRs were also analyzed for SSR-gene association using gene ontology analysis. By utilizing the developing seed transcriptome data within parental lines, the study has enhanced the polymorphic SSR identification precision and facilitated in the RIL population. The undertaken study pioneers the use of transcriptome sequencing and genetic mapping to overcome challenges posed by the intricate wheat genome. This approach offers a cost-effective, less labour-intensive alternative to conventional methods, providing a platform for advancing wheat breeding research.

Charge of a transmembrane peptide alters its interaction with lipid membranes.

Transmembrane (TM) proteins interact closely with the surrounding membrane lipids. Lipids in the vicinity of TM proteins were reported to have hindered mobility, which has been associated with lipids being caught up in the rough surface of the TM domains. These reports, however, neglect one important factor that largely influences the membrane behavior - electrostatics of the TM peptides that are usually positively charged at their cytosolic end. Here, we study on the example of a neutral and a positively charged WALP peptide, how the charge of a TM peptide influences the membrane. We investigate both its dynamics and mechanics by: (i) time dependent fluorescent shift in combination with classical and FRET generalized polarization to evaluate the mobility of lipids at short and long-range distance from the peptide, (ii) atomic force microscopy to observe the mechanical stability of the peptide-containing membranes, and (iii) molecular dynamics simulations to analyze the peptide-lipid interactions. We show that both TM peptides lower lipid mobility in their closest surroundings. The peptides cause lateral heterogeneity in lipid mobility, which in turn prevents free lipid rearrangement and lowers the membrane ability to seal ruptures after mechanical indentations. Introduction of a positive charge to the peptide largely enhances these effects, affecting the whole membrane. We thus highlight that unspecific peptide-lipid interactions, especially the electrostatics, should not be overlooked as they have a great impact on the mechanics and dynamics of the whole membrane.

A Rare Case of Pulmonary Cavitary Disease Caused by Mycobacterium xenopi.

Mycobacterium xenopi is a slow-growing, acid-fast, non-tuberculous mycobacterium (NTM). It is often considered to be a saprophyte or an environmental contaminant. Mycobacterium xenopi has low pathogenicity and is usually seen in patients with pre-existing chronic lung diseases and immunocompromised patients. We present a case of Mycobacterium xenopi causing a cavitary lesion in a patient with chronic obstructive pulmonary disease (COPD) that was discovered incidentally during the low-dose CT scan done for lung cancer screening in a patient with COPD. The initial workup was negative for NTM. An Interventional-guided (IR) core needle biopsy was done given the high suspicion for NTM and revealed a positive culture for Mycobacterium xenopi.  Our case highlights the importance of considering NTM in the differential diagnosis of at-risk patients and pursuing invasive testing if there is a high clinical suspicion.

Continuous manufacturing of monoclonal antibodies: Dynamic control of multiple integrated polishing chromatography steps using BioSMB.

We propose a strategy for automation and control of multi-step polishing chromatography in integrated continuous manufacturing of monoclonal antibodies. The strategy is demonstrated for a multi-step polishing process consisting of cation exchange chromatography in bind-and-elute mode followed by mixed-mode chromatography in flowthrough mode. A BioSMB system with a customized Python control layer is used for automation and scheduling of both the chromatography steps. Further, the BioSMB valve manifold is leveraged for in-line conditioning between the two steps, as tight control of pH and conductivity is essential when operating with multimodal resins because even slight fluctuations in load conditions adversely affect the chromatography performance. The pH and conductivity of the load to the multimodal chromatography columns is consistent, despite the elution gradient of the preceding cation exchange chromatography step. Inputs from the BioSMB pH and conductivity sensors are used for real-time control of the 7 pumps and 240 valves to achieve in-line conditioning inside the BioSMB manifold in a fully automated manner. This is confirmed by showcasing different elution strategies in cation exchange chromatography, including linear gradient, step gradient and process deviations like tubing leakage. In all the above cases, the model was able to maintain the pH and conductivity of multimodal chromatography load within the range of 6 ± 0.1 pH and 7 ± 0.3 mS/cm conductivity. The strategy eliminates the need for using multiple BioSMB units or integrating external pumps, valves, mixers, surge tanks, or sensors between the two steps as is currently the standard approach, thus offering a simple and robust structure for integrating multiple polishing chromatography steps in continuous downstream monoclonal antibody purification trains.

Artificial intelligence and machine learning applications in biopharmaceutical manufacturing.

Artificial intelligence and machine learning (AI-ML) offer vast potential in optimal design, monitoring, and control of biopharmaceutical manufacturing. The driving forces for adoption of AI-ML techniques include the growing global demand for biotherapeutics and the shift toward Industry 4.0, spurring the rise of integrated process platforms and continuous processes that require intelligent, automated supervision. This review summarizes AI-ML applications in biopharmaceutical manufacturing, with a focus on the most used AI-ML algorithms, including multivariate data analysis, artificial neural networks, and reinforcement learning. Perspectives on the future growth of AI-ML applications in the area and the challenges of implementing these techniques at manufacturing scale are also presented.

Tick-Borne Rhabdomyolysis: A Rare Case of Rhabdomyolysis and Acute Kidney Injury Due to Anaplasmosis.

Anaplasmosis is a tick-borne illness commonly seen in the northeastern states of the United States. The most common presenting signs are fever, malaise, and body aches accompanied by leukopenia, thrombocytopenia, and transaminitis. Rhabdomyolysis and acute kidney injury are rare presentations that can lead to significant morbidity.  We present the case of a patient who presented with non-specific symptoms of malaise, fatigue, and body aches and was found to have rhabdomyolysis and acute kidney injury on laboratory workup. A presumptive diagnosis of anaplasmosis was made, and the patient was started on treatment for the same. The patient recovered successfully. Our case highlights the rare presentation of anaplasmosis with rhabdomyolysis and acute kidney injury. Physician awareness is needed for early diagnosis and preventing morbidity.

Process Analytical Technology (PAT) Implementation for Membrane Operations in Continuous Manufacturing of mAbs: Model-Based Control of Single-Pass Tangential Flow Ultrafiltration.

Control of single pass tangential flow ultrafiltration (SPTFF) is crucial for continuous manufacturing of monoclonal antibodies (mAbs). Integrating SPTFF technology into continuous manufacturing trains requires successful resolution of several challenges that arise due to the complexity of mass transfer interactions across multi-membrane configurations, the significant effect of feed material attributes and process variability on flux, and the need for advanced scheduling. In this paper, we propose a real-time, automated monitoring and control strategy for SPTFF in continuous processing of mAbs. The approach leverages a previously developed model for predicting the VCF across an SPTFF module based on the gel polarization model of protein ultrafiltration. A distributed control system (DCS) architecture was created for integrating the monitoring sensors and control elements, including NIRS sensors for concentration monitoring, as well as weighing balances, pressure sensors, pumps, and valves. Two different SPTFF control strategies were developed, firstly for final formulation of the drug product into the drug substance (ultrafiltration and diafiltration), and secondly for in-line concentration between two chromatography steps. Case studies were designed with 15 runs to test the strategy with a range of deviations induced in the feed and process conditions. The retentate concentration was controlled to within 10% of the target value in all runs. The combination of real-time sensor data and model-based control effectively enabled automated and tightly controlled operation of the SPTFF step and is a key enabler of quality by design in continuous mAb manufacturing.

Continuous manufacturing of monoclonal antibodies: Automated downstream control strategy for dynamic handling of titer variations.

Handling long-term dynamic variability in harvest titer is a critical challenge in continuous downstream manufacturing. This challenge is becoming increasingly important with the advent of high-titer clones and modern upstream perfusion processes where the titer can vary significantly across the course of a campaign. In this paper, we present a strategy for real-time, dynamic adjustment of the entire downstream train, including capture chromatography, viral inactivation, depth filtration, polishing chromatography, and single-pass formulation, to accommodate variations in titer from 1-7 g/L. The strategy was tested in real time in a continuous downstream purification process of 36 h duration with induced titer variations. The dynamic control strategy leverages real-time NIR-based concentration sensors in the harvest material to continuously track the titer, integrated with an in-house Python-based control system that operates a BioSMB for carrying out capture and polishing chromatography, as well as a series of pumps and solenoid valves for carrying out viral inactivation and formulation. A set of 9 different methods, corresponding to the different harvest titers have been coded onto the Python controller. The methods have a varying number of chromatography columns (3-6 for Protein A and 2-10 for CEX), designed to ensure proper scheduling and optimize productivity across the entire titer variation space. The approach allows for a wide range of titers to be processed on a single integrated setup without having to change equipment or to re-design each time. The strategy also overcomes a key unexplored challenge in continuous processing, namely hand-shaking the downstream train to upstream conditions with long-term titer variability while maintaining automated operation with high productivity and robustness.

Differences in DNA condensation and release by lysine and arginine homopeptides govern their DNA delivery efficiencies.

Designing of nanocarriers that can efficiently deliver therapeutic DNA payload and allow its smooth intracellular release for transgene expression is still a major constraint. The optimization of DNA nanocarriers requires thorough understanding of the chemical and structural characteristics of the vector-nucleic acid complexes and its correlation with the cellular entry, intracellular state and transfection efficiency. L-lysine and L-arginine based cationic peptides alone or in conjugation with other vectors are known to be putative DNA delivery agents. Here we have used L-lysine and L-arginine homopeptides of three different lengths and probed their DNA condensation and release properties by using a multitude of biophysical techniques including fluorescence spectroscopy, gel electrophoresis and atomic force microscopy. Our results clearly showed that although both lysine and arginine based homopeptides condense DNA via electrostatic interactions, they follow different pattern of DNA condensation and release in vitro. While lysine homopeptides condense DNA to form both monomolecular and multimolecular complexes and show differential release of DNA in vitro depending on the peptide length, arginine homopeptides predominantly form multimolecular complexes and show complete DNA release for all peptide lengths. The cellular uptake of the complexes and their intracellular state (as observed through flow cytometry and fluorescence microscopy) seem to be controlled by the peptide chemistry. The difference in the transfection efficiency of lysine and arginine homopeptides has been rationalized in light of these observations.

Determination of the size of quantum dots by fluorescence spectroscopy.

There has been a lack of quick, simple and reliable methods for determination of nanoparticle size. An investigation of the size of hydrophobic (CdSe) and hydrophilic (CdSe/ZnS) quantum dots was performed by using the maximum position of the corresponding fluorescence spectrum. It has been found that fluorescence spectroscopy is a simple and reliable methodology to estimate the size of both quantum dot types. For a given solution, the homogeneity of the size of quantum dots is correlated to the relationship between the fluorescence maximum position (FMP) and the quantum dot size. This methodology can be extended to the other fluorescent nanoparticles. The employment of evolving factor analysis and multivariate curve resolution-alternating least squares for decomposition of the series of quantum dots fluorescence spectra recorded by a specific measuring procedure reveals the number of quantum dot fractions having different diameters. The size of the quantum dots in a particular group is defined by the FMP of the corresponding component in the decomposed spectrum. These results show that a combination of the fluorescence and appropriate statistical method for decomposition of the emission spectra of nanoparticles may be a quick and trusted method for the screening of the inhomogeneity of their solution.

Process analytical technology in continuous processing: Model-based real time control of pH between capture chromatography and viral inactivation for monoclonal antibody production.

A real time mechanistic model-based control strategy is demonstrated for in-line pH adjustment post-capture chromatography and prior to viral inactivation for continuous processing of monoclonal antibodies. At this point in the process, tight control of pH is essential, as pH fluctuations above 3.5 can result in incomplete viral inactivation, while fluctuations below 3.5 can lead to significant aggregate formation. The present approach predicts the pH profile during the transition phase between chromatography wash and elution steps by modelling the process stream at the column outlet as a mixture of two independent buffer systems. Control of pH in this transition phase is a critical consideration in capture chromatography as a significant amount of mAb material is eluted at this time. The model inputs are buffer concentrations, flow rates, and theoretical pKa values, along with cleaning step conductivity profiles which are readily available from a typical process chromatography equipment. The utilization of the most recent cleaning cycle data as an input to the model allows sensitive calibration to the individual process at hand on a column-to-column basis. The model is able to accurately predict the pH profile throughout the elution, as well as calculate the flow rate of the acid (titrant) required at each time point to maintain the pH consistently at 3.5±0.2. The strategy is demonstrated for various buffers, columns, operating conditions, and process deviations in a three-column continuous process, and is a useful and simple approach for achieving robust control of pH at this critical point in the continuous train.

Near Infrared Spectroscopy as a PAT tool for monitoring and control of protein and excipient concentration in ultrafiltration of highly concentrated antibody formulations.

Excipient concentrations are critical quality attributes of monoclonal antibody (mAb) drug products and affect their safety and efficacy. In manufacturing processes, mAb products are formulated into the buffer containing the desired excipients using ultrafiltration (UF) and diafiltration (DF). Control of excipient concentrations is a challenge during high concentration UF due to electrostatic interactions which lead to excipient concentration drifts. This challenge is of increasing importance due to the growing preference towards high concentration subcutaneous drug formulations over conventional intravenous formulations in the biotherapeutic industry. Excipient concentrations are currently measured using offline RP-HPLC which is time-consuming and not suited for real time control. We propose a novel process analytical technology (PAT) tool for monitoring and control of mAb and excipients in high concentration UF using Near Infrared Spectroscopy (NIRS). The NIRS is able to monitor concentrations within ±1% for mAb and ±2% for two common excipients, L-histidine and acetate. A Python-based controller uses real time concentration data to deliver concentrated excipient stock solutions to the UF reservoir whenever the excipient concentrations drift out of range. The PAT control system is able to achieve the target formulation without manual intervention or at-line analysis and is well-suited for implementation in mAb manufacturing platforms.