Добавил:
kiopkiopkiop18@yandex.ru t.me/Prokururor I Вовсе не секретарь, но почту проверяю Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:
Ординатура / Хирургия / Библиотека им академика М.И. Перельмана / Книга_5387_Библиотеки_им_академика_М_И_Перельмана.pdf
Скачиваний:
0
Добавлен:
15.09.2026
Размер:
16 Мб
Скачать
☆
11 • Systems Biology Approaches 291
Apart from omics tools and techniques, several types of mathematical models are also leveraged for drug target identication and validation. These include disease mod‑ els, biotherapeutic models, dynamic pathway models, and data‑driven statistical learning models. In the disease model, a systems‑level mathematical or statistical framework is developed to describe a given disease in silico. Conventionally, kinetic parameters and rate constants from experiments, literature, and/or informed guess are leveraged to model the disease pathophysiology with ordinary differential equations (ODEs), also known as mechanistic models. Data from multi‑omics such as transcriptomics, proteomics, and metabolomics are also being leveraged to improve the predictability of disease models. Another dimension of complexity in disease models is the interaction of components at multiple scales. This is generally captured by multi‑scale hybrid models such as ABMs (Ji etal. 2017; Menezes etal. 2020). In some cases where data availability is limited to develop quantitative models, a systems‑level qualitative model may be developed that is based on molecular cause‑and‑effect relationships. Table 11.4 captures a few of the recent examples of disease models/metabolic models. For example, Ji etal. developed a hybrid agent‑based model to capture cell‑cell interactions in bone marrow under multiple myeloma conditions (Ji etal. 2017). The model was utilized to predict the treatment effects of three key therapeutic drugs. The study found that the combination of these three drugs
TABLE11.4 Recent examples of studies performed in drug target discovery and target validation of biopharmaceuticals utilizing computational systems biology approach
MODELING APPROACH
Mechanistic Metabolic
Multi-scale
mechanistic
Hybrid ODE
and ABM
Multi-scale and
biophysical
Hybrid ODE
and ABM
DISEASE
MODEL STUDY AND MAJOR OUTCOME REFERENCE
Study performed simulation of adiponectin
disease
Kidney
disease
Cancer
(multiple myeloma)
Cancer
(Glioma)
Tuberculosis Captures various aspects of TB disease and
exocytosis in in response to the reduction of β3ARs observed in adipocytes from animals with obesity-induced diabetes.
Developed a complete nephron model and
tested inhibition of Na 2 (SGLT2) along the proximal convoluted tubule. Results predicted that the segment’s Na+ reabsorption decreased signicantly, resulting in natriuresis and osmotic diuresis.
Developed and applied myeloma growth
model to predict the treatment effects of three key therapeutic drugs. Study found that the combination of these three drugs can potentially suppress the growth of myeloma cells and reactivate the immune response.
The model predicts that cell migration depends
on the relative balance between random motility and strength of chemo-attractants.
predict that biomarkers in the blood may only faithfully represent events in the lung at early time points after infection.
+
-glucose cotransporter
Lövfors
etal. (2021)
Layton and
Layton (2019)
Ji etal.
(2017)
Kim etal.
(2015)
Joslyn etal.
(2022)
(Continued)
292 Biopharmaceutical Informatics
TABLE11.4 (Continued) Recent examples of studies performed in drug target discovery and target validation of biopharmaceuticals utilizing computational systems biology approach
MODELING APPROACH
ABM
host-pathogen model
Mechanistic Diabetic
Mechanistic COVID-19 Model tests the hypothesis that SARS-CoV-2
Machine
learning
DISEASE
MODEL STUDY AND MAJOR OUTCOME REFERENCE
Pneumonia Identied key contributors to alveolar infection
in a mechanistic fashion. Results suggest that S. pneumonia interactions with alveolar epithelial cells contribute to overall infection dynamics as compared to interaction with macrophages.
The study addresses the challenge of kidney disease
Lung cancer Developed machine learning model to explore
heterogeneity for clinical trials of novel anti-inammatory therapies, by modeling how chronic inammation affects kidney function in ve compartments. Such models could be utilized for drug target validation.
infects immune cells and, for this reason, induces high-level productions of inammatory cytokines.
novel antibody sequence space and to accelerate the development of highly potent, drug-like antibodies.
Santos
etal. (2018)
Hofherr
etal. (2022)
Reis etal.
(2021)
Makowski
etal. (2022)
can potentially suppress the growth of myeloma cells and reactivate the immune response. Similarly, Layton and Layton (2019) developed a multi‑scale mechanistic model for neph‑ rons. The model was utilized to test the inhibition of Na+‑glucose cotransporter 2 (SGLT2) along the proximal convoluted tubule. The model predicted that the segment’s Na+ reab‑ sorption decreased signicantly, resulting in natriuresis and osmotic diuresis, thus pro‑ viding insights into potential drug targets for kidney disease. In summary, multi‑omics analyses and mathematical models have been utilized to gain insights into disease models and identify potential drug targets using a systems biology approach.
11.5 APPLICATION OF SYSTEMS
BIOLOGY IN BIOPHARMACEUTICAL
DEVELOPMENT–OPTIMIZING
GROWTH AND PRODUCTIVITY
Once the therapeutic target has been identied and validated, the next step in the bio‑ pharmaceutical pipeline is to produce the therapeutic biomolecule or protein of inter‑ est. This means to develop a scalable process to generate material for various stages
11 • Systems Biology Approaches 293
of studies, including regulatory toxicology studies and phase 1 and phase 2 clinical trials prior to approaching the Food and Drug Administration (FDA) for evaluation (Hu
2020). This set of steps, in the broader sense, can be termed biopharmaceutical develop‑ ment. It encompasses both the discovery and development of the drug target. A typical biopharmaceutical process development paradigm includes developing a cell line that can stably produce the biotherapeutic molecule (cell line development), generating the biomolecule in cell culture (upstream process development), purication of cell culture material to remove impurities (downstream process development), and the nal formu‑ lation of the therapeutic into drug product (drug product development).
The upstream process development of biotherapeutics such as monoclonal antibod‑ ies is typically accomplished using recombinant DNA (rDNA) technology to engineer producer cell lines such as Chinese Hamster Ovary (CHO) cells, mouse myeloma (NS0) cells, among others (Dhara etal. 2018). As such, there are two major goals for biophar‑ maceutical product development: (i) enhance the productivity of the biologics (directly linked to the reduction in the cost of manufacturing) and (ii) enhance and maintain the target range for product quality attributes (directly linked to safety and efcacy of the drug product) (Figure11.4). On both these fronts, systems biology techniques such as multi‑omics analyses and process modeling are implemented to gain process/cellular enhancements.
The productivity of biotherapeutics in cell culture processes directly depends on the amount of cell biomass available and its capability to produce the protein of interest. Cell concentration is generally depicted by viable cell density (VCD). Specic productivity, the rate of protein production per cell per unit time, is depicted by q
. Together, these two
P
key factors determine overall volumetric productivity (VP) as VP is an integral of qP*VCD over the batch duration (time) and vessel volume. Conventionally, bioprocess parameters such as pH, temperature, dissolved oxygen, rate of agitation, etc., are utilized as levers to
FIGURE11.4 Systems approach toward biopharmaceutical development.
294 Biopharmaceutical Informatics
increase growth and productivity. In addition, perturbations in cell culture medium com‑ position, feeding strategy, and perfusion are also leveraged to enhance cell growth and productivity (Ritacco etal. 2018; Hiller etal. 2017). However, these conventional levers have limitations as they cannot be utilized to perturb the cellular physiology from within to push the biological limit for growth and productivity (Hoang etal. 2022). Moreover, these methods require a signicant number of experiments to determine the statistically optimal condition, as they are mostly trial‑and‑error based. The tools and techniques in systems biology can be leveraged to push the limits of growth and productivity by identifying physiological bottlenecks and developing mitigation strategies (Kildegaard etal. 2013). Moreover, as the design of experiments is based on mechanistic insights, it also helps in reducing the number of experiments needed for achieving optimal growth and productivity. Table11.5 summarizes a few recent examples where a systems biology approach has been implemented to achieve improved cell growth and protein productivity. For instance, it is well established that high concentrations of lactate and ammonia result
TABLE11.5 Recent examples of studies performed to enhance cell growth and productivity of biopharmaceuticals using systems approach
TOOLS/ SYSTEMS APPROACH
Multi-omics
and functional analyses
Multi-omics
and functional analyses
Omics and
functional analyses
Multi-omics
and functional analyses
Multi-omics
and functional analyses
TECHNIQUES
EMPLOYED STUDY AND MAJOR OUTCOME REFERENCE
Transcriptomics,
proteomics, metabolomics, and glycomics
Proteomics and
metabolomics
Proteomics The study highlighted key pathways for
Transcriptomics,
proteomics, and metabolomics
Transcriptomics,
and metabolomics
Performed multi-omics proling to
investigate effects of pH on growth and productivity. Study concluded that pH set points differentially regulated various intracellular pathways including vesicular trafcking, cell cycle, and apoptosis, thereby impacting growth and productivity.
Study demonstrated that modications to
cellular environment by modifying feed in cell culture can be guided by omics studies to achieve enhanced cell growth.
targeted engineering to generate desirable CHO cell phenotypes with enhanced growth for biotherapeutic production.
The study identies key metabolic enzymes
and metabolites as indicators of growth and productivity. Also demonstrates that redox balance is key to cellular health during cell culture processes.
The study identies and demonstrates
intermediates and byproducts of amino acid metabolism as secondary growth inhibitors in cell culture for biotherapeutics production.
Lee etal.
(2021)
Blondeel
etal. (2016)
Bryan etal.
(2021)
Ali etal.
(2019)
Chandra
Mulukutla etal. (2017)
(Continued)
11 • Systems Biology Approaches 295
TABLE11.5 (Continued) Recent examples of studies performed to enhance cell growth and productivity of biopharmaceuticals using systems approach
TOOLS/ SYSTEMS APPROACH
Mathematical
modeling
Mathematical
modeling
Mathematical
modeling
Mathematical
modeling
Mathematical
modeling
TECHNIQUES
EMPLOYED STUDY AND MAJOR OUTCOME REFERENCE
Genome-scale
modeling
Genome-scale
modeling
Constraint-based
modeling
Metabolic ux
analysis
Genome-scale
metabolic ux modeling
The study evaluates and compares
parameters for the two phases of cell culture, i.e., growth phase and production phase.
Developed a metabolic network-based
modeling approach utilizing genome-scale model, and implemented it to develop feeding strategies to potential increase protein productivity.
The study identies unconventional
objective functions to minimize non-essential nutrient uptake rate toward cell growth and productivity improvements in various CHO derived cell lines.
The study suggests that productivity is
related to the oxidative state of metabolism whereas cell growth can be characterized by glycolytic metabolic state.
Implemented genome-scale model to
identify feed supplements to enhance biotherapeutic productivity.
Schinn
etal. (2021)
Fouladiha
etal. (2020)
Chen etal.
(2019)
Templeton
etal. (2013)
Huang
etal. (2020)
in growth inhibition in cell cultures, thereby reducing the overall productivity (Pereira etal. 2018). Several process‑related and cell engineering strategies have been devised to reduce the amount of these two growth inhibitors in cell culture. However, even with low concentrations of lactate and ammonia, cells in culture tend to plateau out in terms of growth (Chandra Mulukutla etal. 2017). Mulukutla etal employed omics techniques to identify and quantify several byproducts or intermediates of amino acid metabolism that accumulate in fed‑batch cell culture and impact cell growth. Further, transcriptomics and metabolomics analyses resulted in the identication of cell engineering targets that could reduce the biosynthesis of these growth inhibitors. Metabolic engineering of CHO cells by knocking out such a gene target resulted in signicant improvement in cell growth (Mulukutla etal. 2019). Similarly, mathematical modeling tools within systems biology have also been utilized to enhance protein productivity in producer cells. For instance, Huang etal implemented a genome‑scale model along with transcriptomic and metabo‑ lomic analyses to systematically evaluate CHO cell culture and gain metabolic insights for bioprocess development (Huang etal. 2020). The genome‑scale model was leveraged to identify an experimentally optimal condition for improved productivity. These are a couple of examples that demonstrate how systems biology tools and techniques, includ‑ ing multi‑omics analyses and mathematical modeling, can be leveraged to enhance cell growth and productivity of biotherapeutics in cell culture.
296 Biopharmaceutical Informatics
11.6 APPLICATION OF SYSTEMS
BIOLOGY IN BIOPHARMACEUTICAL
DEVELOPMENT–CONTROLLING
PRODUCT QUALITY ATTRIBUTES
Product quality attributes of biotherapeutics refer to the way the produced molecule exists in the drug substance apart from the amino acid backbone structure. Attributes such as post‑translational modications (glycosylation, deamidation, glycation, etc.), charge variants, fragmentation, and aggregation may be directly linked to the safety and efcacy of the biopharmaceutical drug product. While optimizing cell growth and pro‑ ductivity are the major drivers for developing a cost‑effective biotherapeutics production process, controlling the aspects of the product quality attributes is equally important for developing safe and efcacious biotherapeutics. In cases where the product quality attributes could have a potential impact on drug product safety and/or efcacy, they are termed ‘critical quality attributes’ (CQAs). A well‑dened control strategy is required to maintain a prescribed or clinically tested range of these CQAs during the manufactur‑ ing process (Rathore and Winkle 2009).
Conventionally, the biopharmaceutical development process involves the design of experiment (DoE) studies to identify process parameter ranges that are expected to result in products with CQAs within the prescribed range. However, given the complex‑ ity and heterogeneity within the biological systems, control of CQAs has been a chal‑ lenging task. This is where systems biology tools can be implemented to gain detailed mechanistic insights into the inuence of cell physiology and extracellular environ‑ ment on quality attributes, and how it is impacted by cell culture process parameters (Kildegaard et al. 2013; Sha etal. 2016). As an example, this section will focus on N‑linked glycosylation (N‑glycosylation) to understand how systems biology can be implemented to gain better control and modulation of various glycosylated species dur‑ ing cell culture process development.
N‑glycosylation is a post‑translational modication that impacts various aspects of therapeutic proteins, including effector functions, pharmacokinetic clearance, safety, immunogenicity, stability, and shelf life. Various studies have been performed to understand the effect of cell culture process parameters and media components on N‑glycosylation (Sha etal. 2016). However, to precisely control and modulate the glyco‑ sylated species, a deeper mechanistic insight is required. To this end, both multi‑omics studies as well as mathematical modeling have been utilized to further the understand‑ ing of N‑glycosylation control and modulation. Table 11.6 summarizes a few recent examples where systems biology has been implemented to achieve similar goals of con‑ trolling and modulating post‑translational modications such as N‑glycosylation. For example, Krambeck etal developed a detailed reaction kinetics model for protein gly‑ cosylation to predict the various glycosylated species that can arise in cell culture when enzymes linked to N‑glycosylation are perturbed (Krambeck et al. 2009). By lever‑ aging mathematical modeling and integrating glycomics data from proteins produced
11 • Systems Biology Approaches 297
TABLE11.6 Studies performed to enhance and control product quality attributes of biopharmaceuticals using systems approach
TOOLS/ SYSTEMS APPROACH
Omics and
functional analyses
Omics and
functional analyses
Omics and
functional analyses
Omics and
functional analyses
Omics and
functional analyses
Mathematical
modeling
Mathematical
modeling
Mathematical
modeling
Mathematical
modeling
TECHNIQUES
EMPLOYED STUDY AND MAJOR OUTCOME REFERENCE
Transcriptomics
and metabolomics
Transcriptomics
and metabolomics
Proteomics Differential activation of oxidative
Transcriptomics,
proteomics, and metabolomics
Transcriptomics The study implements omics analysis to
Genome-scale
modeling
Mechanistic,
reaction kinetics model
Mechanistic,
reaction kinetics model
Mechanistic,
reaction kinetics model
The study reveals the effect of oxidative
stress on protein sialylation using multi-omics approach.
The study identies the root cause for
temporal heterogeneity of glycan species in therapeutic proteins produced in cell culture and suggest mitigation strategies based on transcriptomic and metabolomics analyses.
phosphorylation may result in differences in post-translational modication as well as charge variants of therapeutic proteins produced in CHO cell culture.
The study identies key genes and
pathways that are perturbed by pH variation in cell culture, leading to an impact on N-glycosylation, protein aggregation, and charge variant proles apart from mAb productivity.
understand biological causes of aggregation and also identies potential strategies to control this product quality attribute.
Modeling framework to investigate the
effect of ammonium of sialylation of monoclonal antibodies.
Modeled structure-specic turnover
rates of N-glycans, and prediction of N-glycan heterogeneity for glycosylation enzyme inhibitors.
Extension of GLYMMER model to ten
different CHO cell lines with varied activities of enzymes involved in N-glycosylation and NSD transport.
Utilized kinetic model to design and
build 23 different transgenic pools to enhance galactosylation in protein therapeutics.
Lewis etal.
(2016)
Sumit etal.
(2019a)
Strasser etal.
(2021)
Lee etal.
(2021)
Barzadd etal.
(2022)
Savizi etal.
(2021)
Arigoni-Affolter
etal. (2019)
Krambeck etal.
(2017)
Stach etal.
(2019)
(Continued)
298 Biopharmaceutical Informatics
TABLE11.6 (Continued) Studies performed to enhance and control product quality attributes of biopharmaceuticals using systems approach
TOOLS/ SYSTEMS APPROACH
Mathematical
modeling
TECHNIQUES
EMPLOYED STUDY AND MAJOR OUTCOME REFERENCE
Machine
learning and AI tools, glycomics
This study reviews the implementation
of machine learning and AI tools on large set of glycomics data and proposed that such models can be analyzed to gain mechanistic insights into glycosylation machinery and how the machinery shapes glycans under different scenario.
Li etal. (2022)
by a variety of cell lines, they were able to further improve the understanding of gly‑ can heterogeneity in therapeutic proteins (Krambeck etal. 2017). Such models have signicantly helped improve the predictability and controllability of N‑glycosylation. Similarly, Sumit etal. (2019) utilized multi‑omics functional analyses to identify tem‑ poral bottlenecks in the N‑glycosylation pathway during fed‑batch production of thera‑ peutic proteins. The study helped develop mitigation strategies to minimize temporal heterogeneity of glycan species in fed‑batch processes toward better controllability of the system. Mathematical modeling and multi‑omics analyses can also be utilized to identify gene targets that can modulated to precisely control N‑glycosylation in cell culture (Chang etal. 2019). These examples demonstrate that systems biology tools and techniques, including multi‑omics analyses and mathematical modeling, can be lever‑ aged toward precision control of product quality attributes such as N‑glycosylation for developing safe and efcacious biotherapeutics.
11.7 BIG DATA APPROACH IN
BIOLOGICS DRUG DEVELOPMENT
During biopharmaceutical drug development, in particular cell culture process develop‑ ment, a vast amount of time series and discrete datasets are generated. This includes but is not limited to, process data from bioreactors, bioanalytical data such as nutrient, metabolite and byproduct concentrations, cell growth and viability, product titer and pro‑ ductivity, as well as data for product quality attributes. With limited success in phase 1 and phase 2 stages, it is imperative to utilize all the data generated from various programs toward a better process understanding and control (Hay etal. 2014). To this end, digital twins for upstream process development have been conceived and developed (Hoang etal. 2022). Such digital twins utilize hybrid models that incorporate mechanistic under‑ standing from the systems biology approach and statistical learning approach for data with limited mechanistic insights (Macdonald 2022; Park etal. 2021). An overview of
11 • Systems Biology Approaches 299
FIGURE11.5 Digital twins and big data systems approach in biopharmaceutical drug dis­covery and development.
upstream digital twins is shown in Figure 11.5. Briey, we can think of three major levers that can control process performance, improve productivity, and modulate product quality. The rst lever is biological, and it includes the transcriptome, metabolome, pro‑ teome, and redox state of the cells, among many others. The second lever is chemical, and it includes medium composition, nutrient feed composition and feeding strategy, pH, transmembrane equilibrium of macro and micronutrients, etc. The third lever is a physi‑ cal lever, also known as process control, and it includes agitation, power distribution per unit volume, heat transfer, scaling considerations, etc. As all three levers can inuence productivity and product quality, efforts have been made to develop an integrated hybrid model that accounts for all these three levers. Thus, an integral part of the upstream digital twin is the omics state of the cells and its dynamics integrated into a mechanistic or genome‑scale model that is part of a bigger hybrid model incorporating other sets of data. Such digital twins can help in process optimization, predicting course corrections, enhancing process robustness, and expediting development timelines (Park etal. 2021).

11.8 CONCLUSIONS AND FUTURE DIRECTIONS

Biopharmaceutical drug discovery and development is a challenging process owing to complexity the biological systems entail. The systems biology approach offers experimental and computational tools such as multi‑omics analyses and quantitative
300 Biopharmaceutical Informatics
mathematical modeling to dissect such complexities. These tools allow for systems‑level analyses, integrated model development, multiple hypotheses testing, and simulation under various constraints, resulting in increased probability of success during drug discovery and development. The systems biology approach has been applied to drug target identication and validation by analyzing and comparing omics states of control and disease conditions both at the molecular level as well as functional level. In addi‑ tion, systems‑level quantitative disease models have been utilized to dissect underlying mechanisms of regulation. These models also provide a platform for in silico testing of hypotheses for drug target validation. Applications of the systems biology approach in drug development include enhancement of cell growth and productivity of producer cell lines engineered to produce recombinant biopharmaceuticals. This is achieved by utilizing functional analyses and mathematical modeling to identify metabolic bottle‑ necks and developing mitigation strategies either via the cell engineering approach or through media and process changes. Similar approaches have also been successfully implemented to dissect regulations underlying variations in product quality attributes of the biopharmaceuticals and have signicantly aided in the control of these attributes for enhanced safety and efcacy.
With recent advancements in computational capabilities and automation, biophar‑ maceutical drug discovery and development are steering toward the next phase of accel‑ eration to reduce development timelines and bring drugs faster to the patients. To this end, the systems approach (omics and computational modeling) is getting integrated with process analytical technologies (PATs) and advanced sensors and actuators to develop cyber‑physical systems (CPS) that enable big data analysis and real‑time control of process (Rathore etal. 2022). The use of hybrid models and statistical learning tools along with CPS will enable the industry to potentially control and modulate the system at a more precise level with predictive course correction. In summary, recent advances in the big data approach along with the systems biology approach appear to have a very positive outlook and tremendous scope in contributing signicantly to biopharmaceuti‑ cal drug discovery and development. Together with advances in data analytics, systems biology approach can signicantly reduce the cost of drug discovery and development and bring novel biotherapeutics faster to the patients.

REFERENCES

Ali, Amr S., Ravali Raju, Rashmi Kshirsagar, Alexander R. Ivanov, Alan Gilbert, Li Zang, and
Barry L. Karger. 2019. “Multi‑Omics Study on the Impact of Cysteine Feed Level on Cell Viability and MAb Production in a CHO Bioprocess.” Biotechnology Journal 14 (4). doi:10.1002/biot.201800352.
Ang, Joo Ern, Akos Pal, Yasmin J. Asad, Alan T. Henley, Melanie Valenti, Gary Box, Alexis
De haven Brandon, et al. 2017. “Modulation of Plasma Metabolite Biomarkers of the MAPK Pathway with MEK Inhibitor RO4987655: Pharmacodynamic and Predictive Potential in Metastatic Melanoma.” Molecular Cancer Therapeutics 16 (10): 2315–23. doi:10.1158/1535–7163.MCT‑16–0881.