workshop 1

Coffee break

workshop 1

Lunch break

Automation Across Scales: High-Throughput Strain Engineering and Bioprocess Validation of Indigoidine Production in Corynebacterium glutamicum

Biofoundries accelerate the engineering and validation of novel production strains to advance resulting bioprocesses to a technology readiness level ≥ 4. Here, we demonstrate how tiered laboratory automation at Jülich Biofoundry enables end-to-end workflows from in silico design to bioprocess validation, with indigoidine production in Corynebacterium glutamicum as a demonstrator. Initial strains expressing the indigoidine synthetase (BpsA) were generated using semi-automated pipetting. To increase product formation, a combinatorial strain library targeting optimized expression of pathway genes was built using the fully automated AutoBioTech work cell. For high-throughput phenotyping, scalable liquid-handling platforms with integrated cultivation and downstream processing were used to screen 48 strains in parallel with quantitative analytics. Top candidates were evaluated in automated stirred-tank reactors, connected to the same analysis workflow for methodological consistency. As a result, automation enabled the rapid construction and characterization of 61 strains with seven indigoidine producers validated at the 1 L lab-scale. This case study demonstrates that complementary automation infrastructures are essential for combining throughput with industrially relevant process data. Future work will focus on software-driven integration to streamline experiments and support machine learning, strengthening the competitiveness of academic biofoundries.

Autonomous dynamic control of overflow metabolism

Overflow metabolism is a highly undesirable phenomenon that occurs in virtually all cell factories. Although several schemes have been applied to avoid overflow metabolism, this problem remains far from being solved. In this study, a genetic sensor for detecting overflow metabolism was designed, and its substrate specificity and reversibility were demonstrated. The circuit was then used to express the small protein SgrT as an output. SgrT inhibits the main glucose transport system in E. coli, thereby reducing the glucose uptake rate. This strategy eliminated overflow metabolism in batch cultures. To test the performance of the autonomous control, the modified cells were exposed to temporal glucose excess during fed-batch cultures, mimicking environmental fluctuations of large-scale bioreactors. Under these conditions, overflow metabolism was reduced by 80% when the cells were transiently exposed to excess glucose during fed‐batch cultures compared to the control cells. The proposed scheme can be useful for optimizing cellular performance and dealing with heterogeneous conditions typical of large‐scale bioreactors.

Engineering a fully genomically re-coded strain for single-molecule tracking and biomanufacturing

Genomically recoded Escherichia coli (GRE) represents a valuable tool as model organisms for single-molecule tracking via click-labelling. These strains are constantly being engineered to incorporate non-canonical amino acids enabling expanded chemical functionality. However, extensive genome-scale recoding frequently reduces their fitness and imposes metabolic re-wiring limiting the utility of GREs as practical model organisms for live-cell imaging in vivo. This study reports that previously optimized recoded strain was subjected to adaptive laboratory evolution (ALE)2 in rich-defined media (RDM) to recover fitness. The evolved strain displayed improved growth relative to the ancestral GRE6 and a wild-type-like growth confirmed by single-cell analysis relative to the commonly used MG1655. Whole-genome sequencing revealed deletion of the initiation inhibitor ratA. ncAA incorporation was significantly improved relative to the ancestral MG1655 strain. These results point out the importance of applying ALE in parallel to rational genome engineering in expanding the benefits of genomically re-coded organisms, applicable for both fundamental studies as well as improved protein production and biocatalysis.

Engineering cyanobacteria for light-driven oxgygenase biocatalysis on gram scale

Cyanobacteria are ideal hosts for sustainable whole-cell biocatalysis, using sunlight as the sole energy source and eliminating the need for organic carbon substrates. Oxygenic photosynthesis provides both NADPH and O₂ via water oxidation, making these organisms attractive platforms for coupling oxygenases to the photosynthetic electron transport chain.
We engineered Synechocystis sp. PCC 6803 to express a variety of monooxygenases for light-driven oxyfunctionalization of cyclic and linear alkanes. A cytochrome P450-based system enabled aeration-independent hydroxylation of cyclohexane to cyclohexanol by directly utilizing photosynthetically generated O₂. Recent work further elucidated electron transfer from the photosynthetic apparatus to heterologous oxygenases. We also established a two-enzyme cascade comprising a Baeyer–Villiger monooxygenase and a lactonase for light-driven conversion of cyclohexanone to 6-hydroxyhexanoic acid (see Figure). Gram-scale production of this polymer precursor demonstrates the industrial potential of photosynthetic whole-cell catalysis. Current efforts focus on valorizing renewable feedstocks, including terminal hydroxylation of biogenic medium-chain carboxylates.
These studies demonstrate the versatility of light-driven oxygenase reactions in cyanobacteria, identify metabolic bottlenecks and engineering targets, and highlight the potential of glucose-independent photobiotechnology for future net-zero biomanufacturing.

Poster session 1 (Foyer)

Drinks

workshop 2

Coffee break

workshop 2

Lunch break

3D-Printed Agarose TPMS Monoliths as Continuous-Flow Immobilized Enzyme Reactors for Starch Hydrolysis

Conventional Packed-Bed Reactors (PBR) experience significant pressure drops, which limit flow rates. Triply Periodic Minimal Surface (TPMS) monoliths can address this by interconnected channels that maintain relatively low pressure drops with high-viscosity feeds. This work investigates how 3D-printed agarose TPMS monoliths can be used as an immobilized-enzyme reactor (IMER), using α-amylase as a model enzyme.

Monoliths were epoxy-functionalized, and α-amylase was covalently immobilized at ~20 mg per 2 mL of total TPMS volume. The IMER achieved a 35% conversion of 1% starch at a residence time (τ) of 8 min, yielding ~3.5 g/L of maltose at steady-state. Apparent Km decreased with increasing flow rate from 3.77 ± 0.334 mg/mL at τ = 2 min to 0.507 ± 0.083 mg/mL at τ = 1 min (free enzyme: 1.98 mg/mL), consistent with boundary layer thinning and increased mass transfer at higher flow rates. The TPMS monolith processed 7% starch feeds without clogging or structural failure, where PBR would fail. The IMER also reduced the viscosity of 5% starch by 75% after 8 hours of recirculation (τ = 3.2 min and shear rate 1 s⁻¹). The IMER retained 60% and 50% of relative activities after 10 batch cycles and 7 days of continuous use, respectively, confirming its reusability.

Using this as a test case, TPMS-based IMERs suggest strong potential for high-value, low-volume applications such as DNase treatment to reduce viscosity by nucleic acid hydrolysis in pharmaceutical manufacturing.

Alkane Hydroxylation by Anchored Unspecific Peroxygenase

The selective oxidation of methane to methanol with molecular oxygen remains a long-standing challenge in heterogeneous catalysis due to the high stability of its C-H bonds [1]. Interestingly, both of these molecules are essential for a more sustainable chemicals market, as biomethane is a sustainable carbon source while methanol is a key precursor to value-added chemicals [2, 3]. Here, we show that this challenge can be addressed through reaction engineering by using an anchored unspecific peroxygenase from Agrocybe aegerita (rAaeUPO) [4, 5] for alkanes partial hydroxylation in liquid phase, and ultimately to perform the reaction in gas phase.

To the best of our knowledge, this is the first time that the testing and characterizing of the anchored rAaeUPO systems has been carried out for alkanes partial hydroxylation in either liquid or gas phase. The results obtained so far will be used to ultimately focus on the challenging partial oxidation of methane. These findings showcase the bio-catalytic performance as well as the synthetic potential of anchored rAaeUPO for C-H hydroxylation under mild conditions and most importantly, guide the reaction engineering for sustainable alkanes valorization.

[1] M.A. Newton, et al., Chem. Soc. Rev., 2020, 49, 1449-1486.
[2] S. Abanades, et al., Int. J. Environ. Sci. Technol., 2022, 19, 3377-3400.
[3] D. Mahor, et al., ChemSusChem, 2022, 15.
[4] J. Zhang, et al., ChemSusChem, 2026, 19.
[5] X. Han, et al., Nat. Commun., 2024, 15, 1235.

Cascade bio-catalysis for efficient synthesis of nucleosides and nucleotides

Biocatalysis is gaining momentum for synthetic applications in the chemical space of nucleosides and nucleotides. Nucleosides and their 5’-triphosphates are in high demand for emerging technologies in cell-free synthesis, including in vitro transcription. Nucleoside analogues are key components of antiviral, anticancer and RNA-based therapeutics. Multienzyme cascade transformations are increasingly used to prepare nucleosides and nucleotides. Biocatalytic technologies of nucleoside synthesis build on the principle of stereospecific conjugation of an activated pentose moiety to a nucleobase. Phosphorylation of the pentose is often used. Nucleotide synthesis relies on iterative phosphorylation to build the 5’-triphosphate group. The talk will show advances in the synthesis of the C-nucleoside pseudo-uridine and its N1-methyl derivative [1-3]. Efficient approaches to nucleoside phosphorylation will be presented [4].

1 Pfeiffer, M., Ribar, A., Nidetzky, B. (2023) Nat. Commun. 14, 2261
2 Ribar, A., Pfeiffer, M., Nidetzky, B. (2024) Chem Catal. 4, 101127
3 Pfeiffer, M., Krammer, L., Zöhrer, J., Breinbauer, R., Nidetzky, B. (2025) Angew. Chem. Int. Ed. Engl. 64, e202506330
4 Damm, O.T., Pfeiffer, M., Zöhrer, J., Nidetzky, B. (2026) Biotechnol. Bioeng. doi: 10.1002/bit.70247

Developing hybrid living materials utilizing Pseudomonas biofilms for continuous adipic acid production from cyclohexanol

Transitioning to sustainable chemical manufacturing using microbial-driven processes requires developing robust, high-efficiency biocatalysts. We developed and assessed a drip-flow reactor in which recombinant Pseudomonas taiwanensis VLB120 created hybrid living materials (HLMs) by establishing biofilms on a porous poplar–PET material enabling the continuous conversion of cyclohexanol to adipic acid (AA). Biofilm formation over a 7 day period was investigated under in a wide operational parameter space. A multiple linear regression model (MLRM) was used to identify the operational parameters required to achieve high areal biomass concentrations. These parameters were experimentally validated, resulting in a maximum areal biomass of 396 gBDW m⁻². According to the MLRM model, a glucose concentration of 10 g L⁻¹ in M9* medium and a flow rate of 45 µL min⁻¹ enabled the reliable establishment of a biofilm with approximately 300 g(BDW) m⁻² within one week. During the subsequent biotransformation phase, 10 mM cyclohexanol was completely converted to AA over 7 days at a glucose concentration of 2 g L⁻¹. During long-term biotransformation, AA formation remained at an apparent yield of 100% for the first week, even at a glucose feed concentration of only 0.5 g L⁻¹. However, AA production declined during weeks 2 and 3. Our work demonstrates a resource-efficient approach to develop HLMs based on Pseudomonas biofilms to maximize biotransformation performance for AA production.

Poster session 1 (Foyer)

Drinks

workshop 3

Coffee break

workshop 3

Lunch break

Bacterial cellulose–chitosan hybrid hydrogels as sustainable 3D matrix for stem-cell culture

Stem-cell culture commonly relies on Matrigel®, which supports adhesion and expansion but has high cost, batch variability and complex xenogeneic composition. This work developed bacterial cellulose/chitosan (BC/CS) hybrid hydrogels as defined, sustainable and functionalizable alternatives for advanced in vitro culture. BC membranes were produced by Komagataeibacter sucrofermentans DSM 15973 in fructose-rich medium for 12 days, reaching 0.74 ± 0.05 g/L of BC. After purification, membranes were combined ex situ with 1% chitosan solutions of different molecular weights and freeze-dried to obtain 3D scaffolds. Hydrogels were characterized by SEM, EDS, FTIR-ATR and water-retention assays, and pore morphology was quantified using Fiji. Selected scaffolds were functionalized with gelatin through EDC/NHS chemistry to introduce cell-adhesive domains. SEM revealed an interconnected porous network in the BC/CS scaffold (Figure 1), with architecture affected by chitosan molecular weight, confirming tunable microstructure through polymer selection. EDS confirmed chitosan incorporation through nitrogen detection, while FTIR-ATR supported chitosan incorporation and gelatin functionalization through hydroxyl- and amide-related bands. The hydrogels displayed high water retention, relevant for hydrated 3D culture environments. Overall, BC/CS-gelatin biocomposites represent tunable, sustainable and defined alternatives to Matrigel®, supporting integration into upstream bioprocessing workflows.

Controlled Growth Arrest by Sulfate Limitation Enhances Plasmid DNA Production in Vibrio natriegens

After the COVID-19 pandemic, which enabled the first widespread application of mRNA vaccines and accelerated gene therapy development, the global demand for plasmid DNA (pDNA) has increased substantially (Ohlson, 2020). Vibrio natriegens, the fastest-growing known non-pathogenic bacterium, represents a promising alternative production host due to its exceptional growth characteristics and the transferability of established molecular biology tools (Weinstock et al., 2016).
In this study, high-throughput automated cultivations were employed to accelerate the development of a pDNA production process based on V. natriegens. We developed a fed-batch process control strategy that allowed us to switch between distinct metabolic states of the cells. The switch was induced by a sulfate limitation which leads to a controlled growth-arrest and an enhanced pDNA production in Escherichia coli (Gotsmy et al., 2023). By restricting sulfate availability, cellular growth was arrested while substrate uptake and metabolic activity was maintained, promoting a redistribution of cellular resources toward plasmid replication.
The results demonstrate the technical feasibility of pDNA production in V. natriegens and establish the organism as alternative pDNA production host. Furthermore, sulfate limitation was successfully transferred as an effective strategy to improve product formation.

ENGINEERING 3D CELL CULTURE PLATFORMS USING AQUEOUS TWO-PHASE SYSTEMS FOR UPSTREAM BIOPROCESSING APPLICATIONS

Aqueous two-phase systems (ATPS) are emerging as versatile platforms for engineering physiologically relevant three-dimensional (3D) cell culture environments with potential applications in upstream bioprocessing. In this work, we developed reproducible and 3D-ATPS culture platforms based on PEG–dextran systems to support scalable spheroid generation and controlled cell microenvironment engineering using standard laboratory infrastructure.

A systematic workflow was established to optimize polymer composition, droplet volume, and construction strategy; enabling stable phase formation and reproducible culture performance across multiple cell models. Digital image analysis was implemented to quantify droplet morphology, circularity, and stability; supporting data-driven process optimization and platform standardization.

The system was validated using MCF-7 and MDA-MB-231 breast cancer cells, as well as 3T3-L1 adipocytes. Uniform spheroid formation, sustained metabolic activity, and reproducible 3D architectures were achieved across models. The proposed platform demonstrated operational simplicity, scalability potential, and compatibility with high-throughput culture approaches.

These results position 3D-ATPS platforms as enabling technologies for next-generation cell-based bioprocesses, supporting reproducible upstream biomanufacturing workflows for translational research.

Intensified fermentation enables a new dawn for large-scale fermentation

Large-scale fermentative processes face significant restrictions (inhibition, emulsification & fed-batch limitations) that prevent reaching minimum achievable product costs. Continuous processes require cell retention systems that don't scale economically. Hence many fermentative products never reach market at impactful volumes, illustrated by poor investment returns following the 2004 DOE "Top Value Added Chemicals From Biomass" report.
Process intensification through ISPR addresses most of these issues by removing product while retaining cells. The continuous production at fuller fermenter loads for longer periods than fed-batch allows operation near optimum conditions with high yields and rates. Fermentation economics improve dramatically, on cost-of-goods and capex efficiency.
The FAST™ technology enables gravity-based continuous in-situ L/L and L/S separation, improving fermentative production of organic acids, proteins, many chemicals and biofuels. Depending on feedstock, product, and microbe, COG reductions of 20–50% are achievable. Combined with growing demand to reduce fossil dependency, this warrants renewed interest in large-volume biofuels and -chemicals.
The presentation covers continuous separation with cell retention, its scale-up and scale-down, and its positive effects on fermentation, DSP, and overall process economics - illustrated with product examples and results from proprietary lab- and demonstration-scale FAST™ bioreactors.

Poster session 1 (Foyer)

Drinks

Expanding the product range in Synechocystis: stable production of malate in fluctuating environments

Cyanobacterial CO2 conversion holds great promise towards sustainability while challenges remain. One of them is instability of production. When production imposes a fitness burden on the cell, revertants eliminating this burden via spontaneous mutations grow fast, take over the population and impair overall productivity. A strategy to overcome this is via growth-coupled production. The product range of this strategy could be expanded by the concept of relief production, in which a growth-coupled metabolite is converted, resulting in a fitness benefit. In the cyanobacterium Synechocystis sp. PCC6803 it has been hypothesized that relief production of malate can be achieved from the conversion of growth-coupled fumarate. In this case, the enzymes responsible for recycling malate (ME and Mdh) are removed, leading to stable malate production even when overexpressing fumC. Here, we use tightly controlled turbidostats with a small propagation bottle neck (~5%), while applying growth rate as the selection pressure. Results showed stable production in the ΔmeΔmdh and ΔmeΔmdhΔNSI::fumC strains for > 200 generations under continuous light. Furthermore, during dark periods, malate production can also be enhanced by redirecting energy flux generation. Eventually, a full day-night cycle production of malate, as a relief compound, was achieved. Our results suggest that relief production is a promising strategy to extend the list of alternative compounds that can be stably produced.

Implementation of an O2 tolerant hydrogenase in Synechocystis

Coupling of hydrogenases with the cellular metabolism of cyanobacteria bears the potential to fuel biotechnological processes with H2 or to produce H2 via light-driven oxygenic photosynthesis. We introduced the O2-tolerant hydrogenase from C. necator into the cyanobacterial model strain Synechocystis sp. PCC 6803. We characterized the strain Syn_CnSH+ in detail and showed a high hydrogen consumption activity. Additionally, we enhanced activity by adding the native maturation machinery, whereby we achieved a higher level of synthesized protein and elevated activity in vivo. Furthermore, we determined limitations and improved enzyme activity by optimizing physiological conditions, in particular light, substrate (H2), and electron sink (CO2) availabilities. The hydrogen production capacity was improved by co-expressing a transhydrogenase, enhancing the NADH availability in the cells, similar to glucose feeding to Synechocystis. Further, enzyme engineering is needed to capture photosynthetically produced NADPH. Our results give new insights regarding the design of strains for photosynthesis-driven H2 production as well as the can use of H2 to boost biocatalytic reactions or biomass production by phototrophic hosts. Thereby, we aim to enhance sustainable production of chemicals driven by photosynthesis.
Figure 1. Proposed interaction of the soluble hydrogenase of C. necator with the photosynthetic machinery of Synechocystis

Ref.: Lupacchini et al 2021 and Lupacchini 2025

Integration of metabolic flux analysis and process models for mesenchymal stromal cell cultivation

Mesenchymal stromal cells (MSCs) are promising cell therapy sources, while balancing quality and productivity during cultivation remains challenging. Since MSC functions are closely linked to metabolic state, metabolic control has been explored experimentally [1]. However, quantitative design of cultivation processes requires mathematical models that connect culture conditions with dynamic cellular responses. Existing models have mainly described macroscopic culture dynamics [2] or steady-state metabolic fluxes under fixed conditions [3]. In this study, we developed an integrated model that couples metabolic flux analysis with a dynamic process model to simulate time-dependent changes in MSC metabolism and proliferation. The model performance was investigated using experimental data obtained in this study, and the model reproduced oxygen-dependent metabolic flexibility. The glycolytic contribution ratio was newly introduced as a proxy indicator of cellular quality reflecting the balance between glycolytic and oxidative ATP production. Simulations suggested that reducing cultivation time leads to a productivity-quality trade-off, whereas hypoxic conditions can improve the quality-related metabolic state without compromising cell yield. This framework supports a biologically grounded and quantitative basis for MSC cultivation process design.

[1] Burnham et al., Front. Immunol. (2020); [2] Hirono et al., Ind. Eng. Chem. Res. (2022); [3] Fouladiha et al., Cytotechnology (2017).

Scalable Pseudomonas Phenol Bioproduction by Stress Response Engineering

Phenol is a key compound in the chemical industry, serving as a platform chemical for the manufacture of a wide range of products. Since phenol is still predominantly produced from crude oil–based feedstocks, its production represents an ideal example of a conventional chemical process that can be transformed into a more sustainable biotechnological drop-in alternative.

We aim to demonstrate the potential of process-oriented biotechnological research by combining expertise in molecular microbiology with industrially focused process development. To this end, Pseudomonas taiwanensis, as a highly solvent-tolerant microorganism, will be engineered into an efficient phenol production host by specifically modulating its stress response.
The resulting strain will then be directly evaluated and optimized under application-relevant bioprocess conditions through the identification and adjustment of key process parameters. To fully realize the industrial applicability of this approach, the process will subsequently be scaled up to a 2 L lab-scale bioreactor in order to assess enhanced phenol production under industrial operating conditions with second-phase extraction.
We therefore screen several organic solvents for biocompatibility and extraction performance, and apply these in biphasic fermentation. Lastly, the process is run in fed-batch mode with continuous in-situ product recovery via extraction.

Coffee break

A Standardised Microfluidic TFF Platform for Scale Relevant, Low Volume UF/DF Process Development

Tangential flow filtration (TFF) is a core downstream operation for protein concentration and buffer exchange, yet process development is slow and material intensive. Conventional scale down models frequently require large volumes and/or fail to preserve scale defining hydrodynamics, limiting their predictive value. Here, we present a standardised microfluidic TFF (µTFF) platform enabling low volume, scale relevant ultrafiltration and diafiltration operation.
The µTFF device accommodates interchangeable flat sheet membranes, allowing rapid comparison of membrane chemistries and operating conditions. Optimised platform channel geometry preserves wall shear rate and transmembrane pressure, enabling representative UF and DF experiments with feed volumes as low as 10 mL and 1 mL, respectively. Stable operation was achieved over a flux range of 30–1055 L m⁻² h⁻¹, with recoveries above 90% and runtimes of approximately 2.3 hours.
When operated at matched wall shear, the µTFF system reproduced bench scale hollow fibre behaviour, capturing filtration time, TMP evolution, permeate flux trends and membrane dependent performance across different chemistries. Validation with bovine serum albumin and monoclonal antibodies confirmed reproducible buffer exchange and preserved product integrity. By combining low material requirements with scale relevant hydrodynamics, the µTFF platform enables faster and efficient, predictive evaluation of industrial purification strategies and workflows.

Host cell proteome profiling for high-throughput downstream process development of vaccines

In vaccine production, chromatography is crucial to remove host cell proteins (HCPs). Finding optimal process conditions is a resource intensive task task, often requiring multiple experimental tests. Consequently, the biopharmaceutical industry is shifting towards more systematic approaches that increase process understanding while reducing development times. High-throughput process development (HTPD) combines mechanistic models with high-throughput experimentation (HTE). Here, we present a HTPD approach for downstream process development of protein subunit vaccines in E. coli. A mechanistic model that describes the separation behavior of individual HCPs has been established. Its implementation relies on the availability of binding isotherm parameters of HCPs under specific chromatographic conditions. A high-resolution isotherm determination method, which integrates linear gradient experiments with mass-spectrometry proteomics, has also been described. However, transition to high-throughput proteomics measurements is crucial for HTE applications. Current work focuses on optimizing the proteomics workflow to execute the previously developed isotherm determination method in a high-throughput manner. Subsequently, characterization of the adsorption behavior of individual HCPs for multiple chromatographic conditions will be performed. Ultimately, this will enable the creation of a comprehensive HCPs property database to be used in model-based downstream process development.

High-Throughput Screening of Precipitation-Driven Partitioning of AAV Virus-Like Particles in PEG/Phosphate Systems Using UV-Derived Scattering

Virus-like particles (VLPs), including those derived from adeno-associated virus (AAV), are increasingly studied for use in vaccination and gene therapy. While adsorption-based methods remain the method of choice for capture, alternative approaches such as precipitation and aqueous two-phase systems (ATPS) are gaining interest, but require evaluation in terms of yield, selectivity, and reversibility of the precipitate. Their optimization is further limited by the lack of high-throughput analytical methods for monitoring VLP behaviour in complex mixtures.
Here, AAV5-VLPs produced in Escherichia coli were used as a model system. UV-derived light scattering (310–340 nm) was employed as a proxy for particulate content and validated against ELISA. Applied to precipitation screening, the method revealed a clear separation between particulate and soluble biomolecular fractions, enabling rapid assessment of selectivity across different PEG systems.
Low-molecular-weight PEG400 showed enhanced selectivity compared to higher molecular weights (PEG600–PEG1000), retaining up to 89% of the impurities while significantly reducing the particulate signal (13%). The approach was extended to PEG/phosphate ATPS, where precipitation-driven partitioning enabled selective accumulation of VLPs in the interphase while removing impurities. In these systems, PEG400-based conditions maintained low residual particulate signal (11%) and increased the biomolecular fraction retention (up to 97%).

Impact of crystal properties on vacuum filtration of L-tyrosine towards an integrated bioprocess

During the microbial production of L-tyrosine, the amino acid forms needle-like crystals due to its low solubility in the fermentation solution. This in-situ crystallization facilitates partial product purification but presents challenges for its recovery and purification, such as filtration. As the crystal size strongly influences filter cake properties and process performance, it is an important factor in the design of the upstream process of L-tyrosine to facilitate its downstream processing. To examine the relationship between crystal size, pressure and filter cake properties, crystallization experiments were conducted to produce populations with distinct mean lengths (50-250 µm). Then, vacuum filtration experiments were performed at three constant pressures (200-800 mbar). The average specific filter cake resistance increases from 10^7 to 10^10 m.kg^(-1) with decreasing crystal mean length and increasing applied vacuum. This effect was accompanied by a reduction of the filter cake porosity from 0.8 to 0.6. These trends indicate that smaller, needle-like crystals form denser, less permeable filter cakes, particularly under higher vacuum. The pressure dependence further reflects cake compressibility, leading to increased resistance. Combining the findings with upstream insights reveals a trade-off between L-tyrosine productivity and filterability. An integrated model is being developed to identify Pareto-optimal conditions balancing upstream and downstream performance.

Lunch break

Integrated Magnetic Nanoparticle-Based Capture for Intensified Recombinant Protein Production in Escherichia coli

The intensification of upstream and downstream operations is critical to enhance the cost-efficiency and scalability of recombinant protein production. Although continuous and integrated biomanufacturing strategies are well established for mammalian cell systems, comparable approaches for microbial hosts remain underexplored.
Here, we present an integrated production and purification strategy for a recombinant Protein A ligand (rSpA) produced in Escherichia coli. The process combines extracellular secretion with a column-free magnetic capture step using bare iron oxide nanoparticles (BION), enabling direct product isolation from fermentation broth.
Fed-batch cultivation yielded extracellular titers of up to 4 g L⁻¹. Magnetic capture was evaluated both after cell removal and directly in unclarified broth, eliminating centrifugation and column chromatography. At pilot scale, several liters of supernatant were processed using a high-gradient magnetic separator with up to 30 g BION, achieving >90% recovery, >95% purity, and five-fold product concentration. Direct capture from unclarified broth showed comparable binding capacity and purity despite high biomass concentrations, demonstrating robustness in complex matrice.
By coupling protein-secreting E. coli with magnetic nanoparticle-based separation, this study demonstrates a scalable and chromatography-free purification strategy that reduces downstream complexity and enables process intensification in microbial biomanufacturing.

Model-Based Design of Experiments for Spray Drying Using Frozen Airflow CFD-DPM Simulations

Spray dryers are highly dynamic systems in which complex airflow patterns, heat and mass transfer, particle drying, and particle-wall interactions occur simultaneously across multiple length and time scales. Accurately resolving these phenomena using transient Computational Fluid Dynamics (CFD) and Discrete Phase Model (DPM) simulations is computationally expensive. This study presents and validates a frozen airflow CFD-DPM framework for efficient simulation and surrogate model generation of a pilot-scale co-current spray dryer equipped with an external two-fluid atomizer. The workflow combines a database of time-averaged airflow fields with frozen CFD-DPM simulations incorporating particle deposition, hindered drying, and crust formation models. Across four experimental batch cases, the framework predicted measured process outputs with accuracy comparable to conventional transient CFD-DPM simulations while reducing solver time from approximately 15 days to 20 minutes, corresponding to a speedup of about 1000×. The reduced computational cost enabled surrogate model generation using Gaussian Process Regression (GPR) within a Model-Based Design of Experiments workflow. Convergence across all investigated outputs was achieved using 47 frozen CFD-DPM simulations. The resulting surrogate model demonstrated predictive accuracy comparable to the underlying CFD-DPM simulations, enabling rapid design-space exploration and computationally efficient spray dryer process optimization.

Optimizing polyhydroxyalkanoates recovery from mixed microbial cultures using supercritical carbon dioxide coupled with solvent extraction: A Taguchi methodology approach

This research investigates a novel protocol for extracting poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV) from mixed microbial cultures using supercritical carbon dioxide (sCO2) induced cell disruption followed by chloroform extraction (CFE). Using the Taguchi methodology, cell disruption and PHBV recovery were optimized by coupling sCO2 treatment with CFE, resulting in a recovery yield of 94.26 wt.% under optimal conditions of 40 °C, 200 bar, 0.4 mL methanol, and 30 min exposure time. Furthermore, gas chromatography and FTIR analyses confirmed that the extraction method did not alter the chemical composition of the recovered PHBV polymer. The average molecular weight and polydispersity index of the extracted PHBV samples were 108.4 kDa and 2.0, respectively, which were comparable to those of the CF-extracted biopolymer (98.4 kDa and 1.9). X-ray diffraction confirmed the semi-crystallinity of the extracted PHBV, and thermogravimetric analysis and differential scanning calorimetry revealed its thermal stability. The melting point, glass transition temperature, and maximum degradation temperature of the extracted biopolymer were 160.5, −0.3, and 289.5 °C, respectively, indicating a broad range of processability and applications. Thus, this study shows that sCO2 disruption as a pretreatment can be coupled with organic solvent extraction, including green solvents, to create an efficient, environmentally friendly technology for extracting PHA without harmful solvents.

Revisiting Expanded Bed Adsorption: Direct Protein Capture from Crude Broths Using Low-Cost Silica

Expanded bed adsorption (EBA) chromatography offers a unique route to intensify downstream processing by directly capturing target proteins from crude fermentation broths. Despite its early promise, industrial adoption has been hindered by costly materials and operational complexity. Here, we present a simple and cost-effective alternative that revisits EBA using unmodified silica and a minimal affinity tag.
Non-spherical silica particles (63–200 µm) were evaluated as an EBA matrix in combination with an alternating arginine–histidine (HR)₄ tag, enabling electrostatic binding to negatively charged silica surfaces. Stable fluidization was achieved in standard chromatography columns with a threefold bed expansion at 209 cm h⁻¹ (voidage 0.8), allowing direct processing of crude feed without clogging or fouling.
This system enabled single-step purification of tagged proteins directly from crude lysate, achieving purities of up to 97% (SDS-PAGE). The combination of robust hydrodynamics, high selectivity, and low material cost demonstrates that widely available silica can replace specialized EBA media.
These results establish a scalable and economically attractive platform for integrated clarification and capture, highlighting a practical pathway to make EBA-based downstream processing more accessible for industrial applications.

Poster session 2 (Foyer)

Drinks

Dinner (New Church)

Enzymatic treatment of the lipid effluent in a continuous stirred catalytic basket reactor

The present study aimed to develop a treatment strategy for lipid effluents using immobilized lipase for the catalytic hydrolysis of triglycerides from simulated dairy effluent in a continuous stirred catalytic basket reactor (SCBR). A biopolymeric blend composed of microcrystalline cellulose and chitosan (MC/CHI) was employed as a support for the covalent immobilization of Eversa Transform 2.0 lipase using epichlorohydrin as the activating agent. The synthetic dairy effluent was prepared from powdered milk to simulate cheese whey from a small-scale dairy producer in Lorena (São Paulo, Brazil), containing 10 g L⁻¹ of powdered milk solution and 2.7 g L⁻¹ of total lipids. The SCBR system was designed with a maximum reaction volume of 200 mL, including inlet and outlet ports for jacket water circulation and substrate feeding/discharge. Biodegradation assays were conducted using 150 mL of effluent at 55 °C, 1000 rpm, 5% (w/v) biocatalyst loading, and a residence time of 6 h. The immobilized biocatalyst exhibited a hydrolytic activity of 4199.80 ± 89.35 U g⁻¹ and an immobilization yield of 73.06%. During biodegradation, the maximum concentration of released free fatty acids reached 42 mmol L⁻¹ after 8 days, with stabilization observed between 6 and 48 h of operation. These results demonstrate the potential of the developed immobilized system for the treatment of lipid dairy effluents.

Evaluating porous materials for cultivating recombinant Synechocystis PCC6803 biofilms to produce poly-caprolactone monomers from cyclohexanol

Amid growing concern about climate change, sustainable production of chemicals is increasingly important. Phototrophic microorganisms such as cyanobacteria and microalgae fix CO₂ using light and can produce valuable compounds like ε-caprolactone (ε-CL) and 6-hydroxyhexanoic acid (6-HHA), precursors of biodegradable polycaprolactone. Biofilm-based systems, such as catalytic living materials, enable efficient cultivation with higher stability, cell density, and continuous operation, further enhanced by co-cultivation with heterotrophic partners.
This work investigates the growth and biocatalytic performance of mixed-species biofilms formed by the genetically modified strains Pseudomonas taiwanensis VLB120_cdh_chmo and Synechocystis PCC6803_cdh_chmo in a recently developed V-shaped biofilm photobioreactor (DE102026108189.4). Different natural, semi-natural, and synthetic porous materials were evaluated for their ability to support biomass formation and the biotransformation of cyclohexanol (CHOL) to ε-CL and 6-HHA. Continuous biotransformation was conducted over 13 days using four different CHOL concentrations. The results showed that biotransformation efficiency was influenced by the choice of material, while biomass remained comparable across all reactors. For the best-performing material, a maximum productivity of 43 ± 1.6 mmol d⁻¹ m⁻² of ε-CL and 6-HHA was achieved with a dry biomass of 27 ± 2.3 g m⁻².

Photosynthesis-driven whole-cell biocatalysis using rec. Synechocystis sp. PCC 6803 for the production of polymer-building blocks

In this study, Synechocystis sp. PCC 6803 strains were engineered to investigate photosynthetically driven redox biocatalysis using heterologous enzyme cascades from the cyclohexane degradation pathway of Acidovorax, comprising a cytochrome P450 monooxygenase, a cyclohexanol dehydrogenase, a cyclohexanone monooxygenase (CHMO), and a lactonase. The main objective was to use photosynthetic reductants and O₂ to convert substrates such as cyclohexane and cyclohexanol into the polycaprolactone monomers ε-caprolactone (CL) and 6-hydroxyhexanoic acid (6HHA).
Conversion of cyclohexane to CL was achieved with a maximum specific activity of ~10 U/gCDW and stable product formation for 2 hours. Cyclohexanol conversion to CL proved to be more challenging as cyclohexanol strongly inhibits CHMO, even at very low concentrations. Therefore, intracellular CHMO levels were improved to mitigate these inhibitory effects. The best-performing strains achieved specific activities of 10-14 U/gCDW and enabled continuous production of a CL/6HHA mixture from cyclohexanol for 7 days when cultivated as single-species biofilms in a drip-flow bioreactor (DFR). Nevertheless, the direct conversion of cyclohexanol to 6HHA showed superior performance, achieving ~30 U/gCDW. This strain was subsequently evaluated in DFRs for continuous 6HHA production.

Whole-cell Multi-step Biocatalysis for the Renewable Production of 6-Amino hexanoic acid

Multistep whole-cell biocatalysis enables sustainable synthesis of value-added compounds such as polyamide (PA) building blocks from renewable resources. In contrast to fossil-based PA production, requiring energy- and waste-intensive multistage processes, whole-cell biocatalysis employing in-vivo cascades is characterised by highly specific enzymatic reactions under mild conditions without the need for intermediate purification steps. This project focuses on the biosynthesis of 6-aminohexanoic acid (6-AHA), a bifunctional compound with application as polymer building block for Nylon 6, and as a pharmaceutical agent.
We aim to engineer microbial biocatalysts for 6-AHA production from biogenic hexanoic acid or hexanoic acid methyl ester using orthogonal enzymatic cascades. Biocatalytic hosts include Escherichia coli JM101 as well as Synechocystis sp. PCC 6803 for light-driven 6-AHA synthesis.
To analyse and optimise the in-vivo cascade, enzymes were tested individually and in combination. Promising terminal functionalization activities based on monooxygenase, alcohol dehydrogenase and transaminase catalysis have already been achieved. Additionally, alanine recycling via alanine dehydrogenases is investigated to improve in-vivo amino donor supply. The resulting reaction cascade will be implemented in a suitable bioprocess concept, mitigating substrate toxicity, including reactor strategies with substrate feeding controlled by off-gas mass spectrometry and two-phase bioreactors.

Coffee break

A Dynamic 1D Gas–Liquid Modeling Framework for Heterogeneous Bioreactors across Multiple Cultivation Modes

The numerical simulation of heterogeneous gas–liquid bioreactors remains challenging, especially when strong coupling between hydrodynamics, mass transfer, and biological reaction must be considered. To address this issue, we developed a dynamic two-phase 1D model able to describe the spatio-temporal evolution of gas and liquid phases along the reactor height. The framework provides axial profiles of gas volume fraction, dissolved species, and reaction rates, allowing the identification of locally limiting zones in continuous and fed-batch cultivation modes.

The approach is illustrated through two representative bioprocesses: 1) biological methanation where gaseous substrates (H2 and CO2) are converted into CH4, leading to significant changes in gas composition, gas volume fraction and reactor liquid level due to water production by the reaction, 2) aerobic fed-batch cultivation of S. cerevisiae, where changes in feedrate strategy leads to substrate gradients along reactor height as displayed in Figure 1.

The proposed 1D framework is suitable for batch, fed-batch, semi-batch, and continuous cultivation modes, including variable feeding and variable level dynamics. The model offers a computationally efficient tool to study how operating strategies and scale-up impact both local and overall reactor performance. In practice, it enables rapid screening of feeding, aeration, and level-control strategies to identify limiting zones of concentrations along the reactor height.

A Hybrid Oxygen Capture Model Integrated with CFD for Efficient Scale-Up of a Fixed-Bed Bioreactor

Biopharmaceutical industries aim to shorten development timelines while ensuring robust and scalable manufacturing processes. Fixed bed bioreactors (FBRs) are increasingly being utilized for their ability to condense large surface areas into a small footprint, enabling efficient and cost-effective production. However, their geometric characteristics can create scale-dependent limits in oxygen transfer, nutrient supply, and local cell viability. Conventional scale-up approaches often fail to capture interactions between flow, mass transfer, and cellular oxygen demand. In this work, a hybrid cellular response kinetics (CRK) model was coupled with computational fluid dynamics (CFD) and applied to FBR scale-up. Data for model training were generated in a small-scale FBR using adherent HEK293T cells under different oxygen and process conditions. Hybrid CRK model included a feed-forward neural network to estimate oxygen uptake rate and kinetic parameters, which were incorporated into a dynamic mass-balance model and coupled to CFD. The model described local flow and oxygen transport and was used to evaluate oxygen gradients and viability across scales. The calibrated model predicted oxygen profiles, viability, and process trends at larger-scale FBR, demonstrating the accuracy of the hybrid CRK–CFD framework. These results show its potential as a digital scale-up tool for identifying oxygen-transfer bottlenecks and reducing experimental work.

A Low-Cost Programmable Platform for Fermentation Scale-Up Research

Fermentation scale-up, defined here as the transfer of a fermentation process between vessel sizes while maintaining performance, remains difficult because changes in mixing, oxygen and heat transfer, shear, and sensor position can alter microbial growth and process behaviour. Bespoke commercial bioreactors support controlled experimentation; however, their design may limit direct comparability across scale. The work presented in this paper proposes a novel low-cost, programmable fermentation vessel to generate reproducible, real-time datasets for digital twin development, thereby supporting scale-up research. The system uses standard glass vessels with 3D-printed modular lids, and impellers to maintain strict geometric ratios across working volumes, including vessel height-to-diameter, impeller size and probe position. A Raspberry Pi/Python software layer will collect calibrated pH, dissolved oxygen, temperature, and turbidity data, log time-stamped datasets, and provide live visualisation. The system will allow fermentation behaviour across vessel size to be compared and allow the construction of a digital twin focused on data capture and behaviour modelling, with the aim of automating fermentation control. The contribution is an open-source, scalable design that integrates low-cost hardware, real-time sensing, and structured data acquisition to support scale investigations and digital twin development.

A Scalable V-Shaped Photobiofilm Reactor for Limnospira platensis Biomass Production

The cyanobacterium Limnospira platensis has attracted considerable interest due to its high protein content and diverse bioactive compounds, and is now widely used in biotechnology, food supplements, biopharmaceuticals, and cosmetics. Commercial production systems currently rely on cultivating planktonic cells in reactors such as open ponds and tubular photobioreactors. However, these systems often achieve low cell densities, substantially increasing downstream costs associated with harvesting and dewatering. In contrast, biofilm-based cultivation offers a promising alternative for attaining high cell densities while reducing downstream processing requirements.
We developed a scalable V-shaped photobioreactor (V-PBR) that uses porous nonwoven carrier materials mounted on 3D-printed frames arranged in alternating layers and supplied with medium from the top and gas from the bottom. The porous carrier material provides a matrix for cell attachment and biofilm formation, enabling repeated harvesting while retaining embedded cells for biofilm regrowth. Scaling the cultivation area from 260 cm² to 7200 cm² increased areal footprint productivity from 14.6 to 19.7 g m⁻² d⁻¹ despite medium recycling. A 5 m² prototype is currently under development to evaluate industrial scalability. Overall, this work demonstrates that V-PBR is a promising alternative to current technologies for producing highly concentrated, easily harvestable biomass at lower operational costs.

Lunch break

From oxygen transfer to transcriptional robustness: identifying the kLa modulon of Corynebacterium glutamicum

The volumetric oxygen transfer coefficient kLa is among the most widely applied scale-up criterion for microbial bioprocess engineering. However, the kLa can be adjusted through different combinations of e.g. aeration and agitation, raising the question whether identical kLa result in comparable biological states across scales. We used glucose-limited chemostat cultures of C. glutamicum to evaluate the transcriptional relevance of kLa as a scale-up criterion (Figure 1).
To assess oxygen transfer-dependent responses, kLa values of 60, 75 and 90 h⁻¹ were established through defined aeration and agitation combinations, followed by RNA-seq to identify transcriptional profiles associated with altered oxygen transfer conditions. Differential gene expression analyses revealed kLa-dependent transcriptional patterns, indicating that changes in oxygen transfer leave measurable molecular signatures beyond classical key performance indicators such as growth and biomass formation. Based on these data, we identified the kLa modulon of C. glutamicum. From a systems biology perspective, a modulon describes a group of genes that are co-regulated in response to a defined stimulus.
Our methodology provides a systemwide readout for evaluating the biological impact of scale-up-relevant process parameters. Comparative analyses across organisms may identify transcriptional signatures linked to robust bioprocess performance, supporting closer integration of strain and bioprocess engineering.

High-throughput process development using modelling approaches for soluble expression in E. coli

During early stage process development, strains with different molecular biology (MolBio) aspects are screened and scaled-up prior to choosing the strain of choice for expressing the protein of interest (POI). Standardised soluble expression fermentation processes lack transferability between products during strain screening and initial scale up. MolBio aspects play a key role in defining initial ranges for process development, therefore needs to be considered for developing standardized processes. Process models to aid process development approaches usually are from commercial software and state-of-the-art models from academia are seldom used.
In this study, we considered three products and screened 12 BL21 strains differing in MolBio aspects to establish soluble process development workflows. We identified potential synergies between two different protein families during early stage process development with respect to fermentation conditions. We were able to achieve upto 50 % increase in soluble titre concentration using intensified design of experiments (iDoE). Furthermore, physiological characterization workflows were established to enable strain characterization for production of different products.

Model-Based Prediction of Scale-Up Behavior in Engineered E. coli for Enzyme Production: A Coupled Kinetic–CFD Strategy

Mathematical modelling is a crucial tool in the bioprocesses to monitor, control, assess, and optimize the process. In the fermentation processes, models can also be applied to predict the effect of scale change on process performance, thereby enabling seamless process scale-up.
The engineered E. coli M15ΔglyA strain producing mutant NagOX L251R enzyme by fed-batch fermentation process was employed as a model system to develop an unstructured, unsegregated kinetic model. The model was tailored to replicate an experimentally optimized fermentation strategy and validated at lab scale (2 L) through close agreement between simulated and experimental data.
The model’s scalability was evaluated at 50 L fermenter using lab-scale kinetic parameters, where simulated trends showed good agreement with experimental profiles, demonstrating its reliability for scale-up applications.
A computational fluid dynamics (CFD) model was further developed to characterize flow behavior and concentration gradients in 50 L fermenter. Integration of the validated kinetic model with CFD enabled assessment of spatial heterogeneities and their impact on cellular performance. CFD analysis indicated homogeneous distribution of biomass and product, while dissolved oxygen gradients suggested potential oxygen limitation at larger scale.
Overall, this study showed that integrated kinetic–CFD framework provides a robust predictive tool for evaluating performance of scale-up fermentations.

Translating experience of gas-permeable membranes from the wastewater sector to the industrial biotechnology sector

The use of gas-permeable membranes for aeration of aerobic bioprocesses, is emerging as a technology that can deliver benefits. While commercial success has recently been achieved for aeration in full-scale wastewater treatment applications, commercial scale deployment in industrial biotechnology is still lacking. This presentation is based on lessons learned during the successful full-scale deployment of such membranes in the wastewater treatment sector. It will describe and address misconceptions about the use of these membranes and also provide a critical appraisal of their use. The first section of this presentation will cover experiences with membrane materials. Operational experience shows that non-porous polymeric membranes are essential due to (i) durability (ii) the elimination of pore-wetting (iii) adequate permeability of gases, particularly oxygen (iv) comparatively lower cost due to high-volume extrusion processes. Secondly, membrane configuration and module design is discussed. Recent research is uncovering new engineering insights into module design including liquid mixing, fibre tautness/vibration, packing density, bubble interactions and mitigation of water condensation in the membrane lumen. The performance of these membranes in the context of aeration capacity and oxygen flux is discussed.

Poster session 2 (Foyer)

Drinks

Dinner (New Church)

Multifactorial Analysis of HEK293-F Cell Aggregation: Uncovering the Role and Interactions of Bioprocess and Medium Factors.

Cell aggregation often limits viability and transfection efficiency in mammalian cell culture bioprocesses. While the development of suspension-adapted cell lines and specialized media formulations has mitigated these effects, cell aggregation still occurs in both parental and adapted cell lines such as HEK293-F. While several bioprocess and media parameters are known to affect this phenomenon, their interactions remain largely unclear. Additionally, some parameters are expected to influence aggregation, but available data remain limited.
This study addresses this gap by identifying interactions between established parameters, as well as quantifying effects of less-studied parameters. To this end, factorial design experiments will be performed with HEK293-F in 250 ml CSTRs. Interaction studies on established effectors such as hydrodynamics and divalent cations will be followed by screening experiments targeting less-characterized parameters such as temperature and medium ionic strength. Their impact on cell aggregate size distribution will be assessed using particle size analysis. Effects on viability and bioprocess performance will additionally be evaluated. The Kolmogorov length scale will be used as a system-independent metric for hydrodynamic forces. The obtained data will be integrated into a cell aggregation model for improved bioprocess control. This will contribute to mitigation strategies and support optimization of industrial processes.

Overcoming Oxygen Transfer Limitations in High-Viscosity Aerobic Fermentation via Sparger-Integrated Internals and Engineered Bimodal Bubble Dynamics

In aerobic fermentation processes, oxygen emerges as the rate-limiting substrate owing to its low solubility in aqueous broths and further becomes exacerbated in the production of high-molecular-weight biopolymers. The increasing viscosity leads to poor turbulence and increased bubble coalescence, and hence a decay of interfacial area density (IAD) and volumetric oxygen mass transfer coefficient (kLa). Transcending from conventional bioreactor design, a fully 3D-printed bubble column reactor equipped with a novel designed baffled ring sparger (BRS) was developed to address these limitations. According to CFD simulations and bubble distribution analysis, the design effectively causes bubble breakage, disrupts early coalescence, and induces slug flow at the center of the air distributor. As a result, it promotes a bimodal bubble size distribution (BSD), where small bubbles enhance interfacial area density (IAD) for efficient oxygen mass transfer, while large bubbles induce turbulence and bulk circulation to improve mixing. By decoupling mixing and oxygen mass transfer, the BRS shows superior measured kLa values and improved performance in high-viscosity fermentation, leading to higher biomass accumulation (40%), final product titer (87%), and volumetric productivity (40%).

Particle-assisted cultivation of filamentous fungi in shaken bioreactors for controlling morphology

Particle-supported cultivation of filamentous fungi allows for the production of different morphologies during a seed-train process, which affects growth and productivity. While the concept has recently been applied to cultivations in rocking-motion bioreactors for the ascomycete Aspergillus niger [1] and in co-cultures together with the filamentous bacterium Streptomyces coelicolor [2], it has not yet been applied to basidiomycetes like Fomes fomentarius. This fungus is used for the production of biomaterials.
We demonstrate how particles of varying composition, size, and concentration can be used to achieve specific morphologies and influence growth at the well-plate and shake-flask scale. This concept can be used to achieve specific surface colonization properties for biomaterial production. It allows for targeted morphology control ranging from dense pellets to scattered growth on different substrates. The investigations are combined with computer-assisted image recognition and enable the correlation of specific structures with adhesion and growth characteristics. The resulting insights can be used to achieve specific material properties across scales.

Literature
[1] Kheirkhah, T., Neubauer, P., Junne, S. (2023), Biochem Eng J, DOI:10.1016/j.bej.2023.108905
[2] Engelbert, K., Kheirkhah, T., Deffur, C. et al. (2026), Appl Microbiol Biotechnol, DOI:10.1007/s00253-026-13822-0

Shedding light on metabolic burden in (industrial) bioprocesses: Quantifying cellular stress across expression levels in Escherichia coli

Metabolic burden is a central challenge in upstream processing, where high‑level recombinant protein production often compromises host physiology and limits process efficiency. Genetic manipulations such as plasmid‑based overexpression disrupt tightly regulated cellular networks, forcing heterologous pathways to compete with native processes for limited resources. This impairs biomass formation, product yields, and bioprocess robustness. We develop a multiple-fluorescent reporter system in Escherichia coli BL21(DE3) to monitor and quantify metabolic burden during heterologous gene expression at transcriptional, translational, and post-translational levels. A constitutive mTagBFP2–F30‑broccoli module reports translational and transcriptional stress via blue and green fluorescence, respectively. To capture post-translational burden, red fluorescent protein mScarlet-I3 is fused to a ClpXP promoter sequence. The ClpXP protease complex, central to protein quality control and the RpoS-mediated stress response, is strongly induced under high proteolytic demand, thereby enabling detection of protein folding machinery overload, a common symptom of metabolic burden. Monitored with automated real-time flow cytometry during bioprocesses, this strain enables real‑time single‑cell resolution of burden dynamics. Overall, this approach aims to provide a novel tool to improve the development of efficient bioprocesses and robust microbial production systems.

Coffe break

The Fundamentals of Bioprocessing for Integration of Tyrosine Production

Bioprocesses are one of several ways of producing chemicals, and because of their potentially lower environmental impact, they are of growing interest. The global demand for sustainable alternatives creates a robust market for biobased chemical production. However, developing robust bioprocesses requires comprehensive insight into both upstream and downstream operations and, critically, into their interdependencies. Key upstream challenges include strain selection, media composition, hydrodynamic conditions, mass transfer, environmental parameters such as pH and temperature, and the rheological behavior of the fermentation broth. Downstream considerations involve cell harvesting, product recovery, purification, and yield loss at each processing step. The integration of both stages presents additional complexity, as upstream titer, rate, and yield (TRY) directly govern downstream operability, purification load, and overall process economics. A structured decision framework is proposed to map critical upstream constraints to broth viscosity, cell density, and TRY performance onto downstream process selection, ensuring that early-stage bioprocess decisions are made with full awareness of their downstream consequences. Taking tyrosine production from E. coli as a case example, the low aqueous solubility and crystallization sensitivity of tyrosine make this constraint-driven decision approach particularly valuable for guiding scalable and economically viable process development.

Thirteen in One: Production and Purification of a Multi-Epitope Vaccine Candidate Against Schistosomiasis

Schistosomiasis is a major neglected tropical disease affecting over 240 million people worldwide. Praziquantel remains the main treatment, but reinfection is common, underscoring the need for effective prophylactic strategies. Vaccine development is challenging because helminths possess sophisticated mechanisms that modulate host immune responses and promote parasite survival. Our group previously identified Schistosoma mansoni peptide epitopes recognized by antibodies from self-curing rhesus macaques using phage display and peptide microarray analyses. Based on these findings, thirteen epitopes were assembled into a chimeric construct, generating the vaccine candidate PolySchisto-Rhiz. Here, we report the rational design, production, and purification of this recombinant protein. Computational analyses guided construct optimization, improving predicted stability and immunogenicity. The optimized sequence enabled an Escherichia coli expression system capable of producing 141 mg/L of recombinant protein under non-optimized conditions in a pilot-scale batch bioreactor. Purification was performed using a two-step immobilized metal affinity chromatography protocol, yielding the target protein with approximately 90% purity and 65% recovery. Together, these results establish a scalable production platform for PolySchisto-Rhiz and provide material for future immunological studies to evaluate its potential as a schistosomiasis vaccine candidate.

Toward sustainable production of bacteriocins from lactic acid bacteria: medium optimization, fed-batch fermentation and membrane-based recovery

The increasing demand for natural antimicrobial compounds has renewed interest in bacteriocins produced by lactic acid bacteria (LAB) as sustainable alternatives to chemical preservatives and conventional antibiotics. Due to their antimicrobial activity, safety profile, and ability to modulate microbial communities, LAB-derived bacteriocins are promising candidates for applications in healthcare and food preservation. However, large-scale production is limited by the high cost of conventional culture media and the complexity of downstream purification processes, which often rely on multistep, strain-specific methods with low yields.

In this study, we developed an integrated and scalable strategy for bacteriocin production and recovery from four LAB strains. Culture media were optimized using a design-of-experiments approach to enhance microbial growth and bacteriocin synthesis. Batch and fed-batch fermentations were compared to identify the most productive conditions. Fed-batch cultivation significantly improved process performance, resulting in nearly threefold higher biomass and protein production than batch cultures.

For downstream processing, membrane-based separation systems were employed to selectively concentrate bacteriocins, providing a more sustainable alternative to conventional purification methods. The resulting bacteriocin-rich fraction showed strong antimicrobial activity against relevant intestinal and skin pathogens.

Towards a platform process for recombinant protein production in Vibrio natriegens

Vibrio natriegens has emerged as a promising host for recombinant protein production due to its exceptionally high growth rate, high protein synthesis capacity, and the recent expansion of its genetic toolbox. Despite these promising characteristics, recombinant protein production has predominantly been done in shake flasks to date. Hence, key challenges in bioprocessing, such as high oxygen demands at higher cell densities, the high salt demand in the media and sensitivity to overflow metabolism and by-product formation remain largely unexplored. To achieve industrially relevant product titers with V. natriegens, it is necessary to establish a scalable platform process and to identify critical process parameters that enable robust growth and high productivity.
The aim of this work was to develop an upstream process comprising batch, fed-batch and induced fed-batch phases, which could serve as a basis for further upscaling of V. natriegens bioprocesses. Therefore, different critical process parameters were identified and analyzed. For this purpose, cultivations were conducted in bench-scale stirred-tank bioreactors with real-time monitoring of key process parameters, including temperature, pH, dissolved oxygen, and off-gas composition. Complementary offline analyses of biomass, metabolites, and product titers enabled a detailed characterization of V. natriegens in a bioreactor setup. This works marks the first steps towards scalable protein production with V. natriegens.

Lunch break

Upstream Process Development for Continuous pDNA manufacturing: A combined experimental and modelling approach

The rapid expansion of cell and gene therapies, mRNA and DNA vaccines has transformed plasmid DNA (pDNA) into a valuable commodity for the biopharmaceutical industry. To meet this surging demand, production paradigms are rapidly evolving to industrialized, GMP-compliant processes. Despite these strides, critical bottlenecks persist across the entire bioprocess train. In the present study, E. coli DH5α was employed as a host strain for the synthesis of supercoiled pDNA under different operating modes and aeration strategies. Initially, a total of 10 bioreactor campaigns were run in batch (8) or fed-batch mode (2). The objective was to map the influence of Critical Process Parameters (CPPs) on bacterial biomass accumulation and pDNA titer. Initial analysis revealed highly non-linear interaction dynamics between aeration rates, impeller speed, specific growth rate, and pDNA productivity. To enhance process understanding, a hybrid (grey-box) model was trained on a combination of the in-house developed and literature derived data. Subsequently, a progressive transfer learning framework was employed to adapt the pretrained model from batch/fed-batch to continuous cultures. This approach exploited historical data to substantially reduce model recalibration costs, emphasizing the potential of hybrid and transfer learning methods to accelerate pDNA bioprocess development. The recalibrated model was used to identify CPP values that maximize pDNA productivity in continuous fermentations

Upstream Bioprocessing Strategies for the Co-production of Omega-3 Fatty Acids and Fucoxanthin by Phaeodactylum tricornutum

Phaeodactylum tricornutum is a marine diatom which produces both eicosapentaenoic acid (EPA, an omega-3 fatty acid) and fucoxanthin, a valuable brown carotenoid pigment with antioxidant properties and claimed anti-obesity health benefits. A small number of researchers have focused on the simultaneous production of both compounds. However, the reported titres and specific concentrations are well below those of the compounds when optimised individually in our laboratory, where it was found that fucoxanthin production was favoured by low light and high nitrate conditions, while the specific EPA concentration remained largely constant across conditions, making volumetric productivity the primary driver of yield.

In this work we have investigated the effects of salinity, CO² concentration, nitrogen availability, light intensity and light quality on the co-production of EPA and fucoxanthin. The favourable conditions identified previously indicate that a single regime maximising the production of both compounds is achievable. Preliminary results demonstrate volumetric fucoxanthin concentrations approaching 80% of those achieved under individual optimisation in our laboratory, with specific EPA concentrations of 53 mg/g, consistent with our group’s earlier findings. Together these will identify the conditions that favour the production of both compounds, informing upstream production strategies for a high-value biorefinery.

Vibrio natriegens as a novel host for RNA therapeutics

Vibrio natriegens, discovered in 1958, is a marine Gram-negative bacterium with the fastest growth rate known to date among living organisms. This fact and its metabolic versatility make V. natriegens as a potential new microbial chassis for biotechnology applications like plasmid DNA (pDNA) production or recombinant proteins expression. The safety of this bacterium has already been proven, and it is classified as a biosafety level 1 organism. Taken together, these factors position V. natriegens as a promising candidate for cell host in various biotechnology fields. However, despite having been discovered in the 50’s, there is a huge knowledge gap about this bacterium, specially in terms of pDNA production yields that can be achieved. In recent years, pDNA market has grown exponentially principally due to their applications in DNA vaccines or in gene therapies. So, the production capacity should be able to keep pace with demand. With this, the aim of this work was to access the capabilities of V. natriegens to produce pDNA intended for mRNA and siRNA production and compare it with Escherichia coli to evaluate whether its rapid growth can be leveraged to improve pDNA production efficiency. The results shown that it is possible to take advantage of the rapid growth of V. natriegens to the production of higher amounts of pDNA in less period of time. It was also possible to confirm that the pDNA produced by V. natriegens was as functional as the one produced by E. coli.

pH control and ethanol-to-acetate ratio as key levers for n-caproate selectivity in pure Clostridium kluyveri batch cultures

Microbial chain elongation via reverse β-oxidation converts ethanol and acetate into medium-chain fatty acids, with Clostridium kluyveri as the model organism. Controlled pH and the ethanol-to-acetate (E:A) molar ratio jointly determine whether elongation is directed toward n-butyrate (C4) or n-caproate (C6). Eleven batch fermentations with C. kluyveri DSM 555 were run in 3-L bioreactors (1.5 L working volume, 37°C, N₂/CO₂ atmosphere) across five pH conditions (uncontrolled ~5.9–6.5, and controlled at 6.2, 6.5, 6.8, 7.1) and initial E:A ratios of 1.2–4.9. pH 6.5 with E:A ≈ 4.8 maximized n-caproate (182.6 mM; 21.2 g/L) with C6 selectivity of 91.5% at the production maximum. C6 selectivity remained ≥90% in the range pH 6.5–7.1 whenever E:A > 3.5, whereas high E:A at pH 6.2 limited selectivity at ~64–80%, indicating that adequate pH is also required. Uncontrolled-pH cultures gave the highest volumetric productivity (1.65 mM/h) but lower titres (~81–115 mM, figure 1), revealing a trade-off between production rate and total accumulation. Two independent experiments at E:A ~ 1.2 (pH 6.2 and pH 6.8) both inverted selectivity toward n-butyrate (C6 ~ 41–44%), showing that the donor-to-acceptor ratio acts as a metabolic limitating independent of pH control. These results define quantitative operating windows for C. kluyveri in stirred-tank systems and provide a basis for bioprocess design in caproate-targeted biorefineries.

Poster session 2 (Foyer)

Drinks

Dinner (New Church)

Applications of Raman Spectroscopy and Mid-Infrared Analytics for Characterization of T-Cell Phenotype and Metabolic State

Raman and mid-infrared (Mid-IR) spectroscopy were evaluated as non-destructive, at-line tools for T-cell culture characterization. Primary human T cells were isolated via immunomagnetic separation (Miltenyi AutoMACS Neo), recombined at five CD4:CD8 ratios (90:10 to 10:90), and expanded over a five-day static culture. Raman (MarqMetrix) and Mid-IR (Sartorius CLADE) spectra were collected alongside BioHT & Metaflex metabolite data. Spectra were preprocessed using Savitzky-Golay first-derivative transformation and modeled via PLS regression in SIMCA.

Both platforms yielded comparable performance at three PLS components. Strong cross-validated predictions (Q²Y>0.65) were observed for pyruvate (Mid-IR 0.91; Raman 0.84), phosphate (Mid-IR 0.91; Raman 0.93), glucose (Mid-IR 0.68; Raman 0.72), ammonia (Raman 0.92), pH (Mid-IR 0.67; Raman 0.73), potassium (Mid-IR 0.83; Raman 0.73), and calcium (Mid-IR 0.82; Raman 0.61), suggesting strong predictive monitoring of cell culture health. Several analytes lacked spectral activity or concentration range showed weak predictions (Q²Y

Automated PAT Model Development Using Genetic Algorithm-Based Pipeline Optimization

For calibration, Process Analytical Technology (PAT) frameworks commonly combine spectroscopic sensors with offline analytical measurements to enable real-time monitoring of bioprocesses. This coupling is typically achieved using chemometric or machine learning models such as Partial Least Squares (PLS) regression. However, model performance strongly depends on appropriate data preprocessing, often requiring the combination of multiple mathematical transformations into preprocessing pipelines. Due to the large number of available preprocessing operations and the varying requirements across datasets and sensor types, selecting suitable preprocessing strategies and model configurations remains a complex and expertise-driven task.
In this study, we present an automated framework for PAT model development based on a genetic algorithm. Instead of manually selecting and optimizing preprocessing steps, hyperparameters, features, and predictive models, the proposed approach autonomously identifies optimized model configurations without user intervention. The framework was evaluated across multiple sensor types (Raman, infrared, UV/Vis) and protein systems (antibody-drug conjugates, virus-like particles, and lysozyme) using datasets from previously published studies, demonstrating competitive predictive performance. Overall, the proposed method simplifies PAT model development workflows and has the potential to substantially accelerate model generation and deployment.

Balancing Prediction Performance and Spectral Interpretability: Multi-Metric Guided Design of Preprocessing Pipelines for Rapid Calibration of Raman Spectroscopy Models.

Raman spectroscopy is a widely explored in-line process analytical technology due to its high chemical specificity and suitability for chemometric modelling. However, calibrating the quantitative models using bioprocess spectra is costly and time-consuming. Single-compound spectra offer a rapid alternative, where signal preprocessing is key to highlight features relevant for transfer to bioprocess data. In this work, a multi-metric framework for systematic selection of spectral preprocessing pipelines will be presented, screening ~65,000 preprocessing combinations, where each pipeline was evaluated across signal preservation, noise reduction, peak resolution, and model performance. Preprocessing pipelines selected with the multi-metric framework were used to calibrate Partial Least Squares (PLS) models on single-compound spectra and predict ethanol and glucose concentrations in four aerobic Saccharomyces cerevisiae batch fermentation datasets. Compared to selection based solely on prediction error of process data, multi-metric selection yielded robust, chemically interpretable, and transferable models. Despite higher glucose errors, ethanol predictions were comparable to literature values, but without using an internal standard (peak-based normalization) and with fewer PLS latent variables. This shows that preprocessing design is a multi-objective problem, balancing prediction performance and chemical interpretability for meaningful calibration transfer.

Multimodal spectroscopy-based real-time monitoring and anomaly detection in ADC manufacturing process

Antibody–drug conjugates (ADCs) represent a promising class of cancer therapeutics that require precise control during manufacturing, particularly during the cysteine-based reduction process, where interchain disulfide bonds are cleaved to expose reactive sites for drug conjugation. Current monitoring approaches rely on offline analytical tools such as HPLC, which provide delayed feedback and fail to capture real-time process dynamics, thereby limiting process optimization and quality control capabilities and ultimately impacting therapeutic outcomes such as drug-to-antibody ratio (DAR).
This study focuses on the development of an integrated soft-sensor-based multimodal spectroscopic monitoring system combining Raman and Fourier Transform Infrared (FTIR) spectroscopy with artificial intelligence and machine learning algorithms for real-time monitoring and anomaly detection during ADC reduction processes. A multimodal spectroscopy approach was employed to capture complementary molecular information throughout the reduction process. Spectral data from both modalities were integrated using advanced data fusion techniques and analyzed using machine learning models, including autoencoder neural networks for unsupervised anomaly detection. This work establishes a foundation for next-generation process analytical technology (PAT) in ADC manufacturing.

Coffee break

Predictive control and real-time optimization is a necessity for robust and high performing bioprocesses: experimental emphasis from P. chrysogenum, CHO, C. glutamicum and E. coli

Microorganisms are the workhorses for modern sustainable production. In a bioprocess the catalytic unit which converts the substrate to a product is a living cell. In the conversion from substrate to product many reactions occur inside the cell, which are highly dependent on the environment (e.g. nutrient concentration, temperature, pH, Ionic strength etc.). In Bioreactors these environmental factors are controlled at optimal conditions, but any deviation can lead to irreversible changes which require stabilizing actions. Classical reactive control is often too late as the culture is already irreversibly damaged, therefore predictive controllers like MPC are needed to plan preventative action ahead. By nature, microorganisms will adapt or degenerate over time which means that optimal condition planned ahead will change and re-optimization of set-points is needed.
Based on different experimental case studies this contribution will show how predictive control and real-time optimization can be used to guarantee stable and high performing bioprocesses. The case studies will include results from P. chrysogenum, CHO, C. glutamicum and E. coli from research involved in the last 10 years.

Scalable, continuous production of mRNA with real-time prediction of titer and purity via a Digital Twin

The rapid development of mRNA-based vaccines and therapeutics has exposed key limitations of conventional batch manufacturing, including low throughput, high costs, and limited scalability. To address these challenges, this study presents an integrated, continuous manufacturing approach combining kinetic modeling, process optimization, and advanced reactor design within a unified platform.
A mechanistic model of the in vitro transcription (IVT) process was developed, incorporating impurity generation. The model was validated and a model-based optimization of reaction parameters resulted in a 55% increase in mRNA yield.
The optimized process was successfully transferred to a segmented plug flow reactor, enabling continuous operation and supporting high-throughput screening and production within the same equipment, achieving a 10-fold increase in productivity and allowing rapid evaluation of up to 20 vaccine candidates. A key contribution is the qualification of a single modular plant capable of operating across scales, from ~1,000 clinical doses to 10 million commercial doses. Process analytical technology enables automated control and precise product slug tracking, reducing product loss and limiting contamination to

Subpopulation-based bioprocess design and control

Using automated flow cytometry coupled to continuous cultivation, we monitored real-time subpopulation (SP) dynamics across microbial systems and showed that SP emergence tracks critical physiological transitions, making population structure a direct source of process information. In P. putida KT2440 grown on benzoate fluorescent biosensors quantified cellular responsiveness, the fraction of cells able to activate a target gene circuit after an environmental shift. Benzoate reduced responsiveness concentration-dependently, independently of the specific circuit. Unlike classical phenotypic switching costs, this impairment occurred without increased population entropy and correlated strongly with fitness loss. A resource-allocation model explained the mechanism i.e., benzoate creates a three-way competition for limited cellular resources and, above a critical burden, insufficient resources remain for adaptive gene-circuit activation. In continuous culture, accumulation of a non-responsive SP emerged as an early warning signal preceding population collapse. This indicator was exploited in a two-stage interconnected bioreactor where benzoate feeding was autonomously regulated from real-time population structure measurements, achieving complete substrate consumption at otherwise destabilizing concentrations. Feedback control grounded in population structure, rather than bulk physiology alone, offers a robust framework for autonomous biomanufacturing.

Establishment and Digital Monitoring of Quillaja saponaria Plant Cell Suspension Cultures for QS-21 Production

Plant cell suspension cultures offer a flexible and contained platform for the production of high value plant secondary metabolites, yet process development is often hampered by limited process understanding and a lack of real time monitoring tools. In this work, a Quillaja saponaria plant cell culture (PCF) line was established and characterized with respect to QS 21 production and biomass growth in a 2 L stirred tank bioreactor.
Process analytical technologies were implemented to enable real time prediction of substrate consumption, biomass formation, and QS 21 production, providing a basis for data driven process monitoring and control. Complementarily, a reduced metabolomics model was developed to describe and predict biomass growth and product formation under varying process conditions.
The combination of PAT based soft sensors and a reduced kinetic model offers a powerful toolbox for the systematic characterization of Q. saponaria PCF processes. This framework paves the way towards model assisted, digitally supported manufacturing of QS 21 and related plant derived biopharmaceuticals.

Lunch break

Avatar-based Conversational AI Assistant in Chemical Engineering Education

The rapid adoption of conversational artificial intelligence (AI) in higher education, which are technologies that allow users to interact with computers using natural language, is reshaping established teaching models and student support systems. Although the enthusiasm for these tools has been spread among both teachers and students, critical gaps remain regarding the measurable impact of these tools on deep learning and academic performance.
This study examines the introduction of a conversational AI avatar to a M.Sc. course taught at the Technical University of Denmark (DTU). Students are divided in three groups to perform an interview-like role-play exercise, where the entity providing answers can be (i) a physical teacher, (ii) an AI-powered chatbot, and (iii) the AI-based conversational assistant represented by the avatar.
Using a mixed-methods design, we evaluated academic performance alongside qualitative student surveys. The results showed that students in all groups are overall satisfied of their experience and received high-quality answers. From a qualitative perspective, students highlighted their support towards the initiative, which aims to raise the enrolment limit in the course and allowed a 66.6% increase in student uptake.
The findings suggest that conversational AI can function as a scalable tool to enhance student engagement, but its adoption must be accompanied by strong institutional policies to address technical inaccuracies.

Closing the Gap: Challenge-Based, Industry co-created laboratory for undergraduate bioprocess development

Modern biomanufacturing demands graduates who can integrate Quality-by-Design (QbD) principles, process analytical technology (PAT) and data analytics to efficiently design and control a bioprocess. This work describes the development of an “Industrial Bioprocess Development laboratory”, a final-year module within the BSc in Bioprocessing at Dublin City University (DCU). This module was co-developed with Sartorius as part of the DCU Futures initiative, a strategic framework to create Challenge-based, industry-partnered learning and transversal skill development.
Students were challenged to maximise the biomass concentration of lactic acid bacteria (LAB) for an industrial functional-food application, providing an authentic and industrially relevant context. Using a challenge-based learning (CBL) framework, students progressed from Design of Experiments tutorials, through to MODDE-guided media design to hands-on batch and fed-batch cultivation with in-line PAT monitoring, offline analytics and software-supported multivariate data analysis using SIMCA.
Sartorius co-development shaped both the tools and workflow: students operated bench-scale BIOSTAT A bioreactors, applied industrial-grade DoE and PAT software, and engaged with academic and industry experts during delivery. Assessment combined practical evaluation, scientific reporting, group presentations and reflective journalling, aligning technical bioprocessing competencies with digital literacy and professional communicatio

Helix Lab - Center for Applied Research and Education in an Industrial Environment

The transition towards data-driven and sustainable biomanufacturing requires professionals who can work across disciplines, organizations, and industrial boundaries. In Kalundborg, Denmark, a unique education and research environment has emerged within one of the world's leading industrial symbiosis ecosystems. Helix Lab was established as a facilitator connecting students, researchers, and industry, transforming industrial challenges into collaborative research and learning opportunities.

The Helix Lab model creates the conditions for applied research and innovation in an industrial setting. Through MSc and PhD projects, Applied Research Seminars, and state-of-the-art laboratory facilities, students and researchers engage with real-world challenges while companies gain access to academic competencies and future talent.

Five years into the initiative, the research portfolio reflects the needs and opportunities of the Kalundborg ecosystem. Current focus areas include advanced process monitoring and control, applied robotics, optimization of resource and energy flows across co-located companies, and circular water solutions.

This contribution presents the Helix Lab collaboration model and discusses how Kalundborg's industrial, educational, and research ecosystem functions as a living laboratory. It further explores how such an environment can accelerate the transition towards Bioproduction 4.0 while creating opportunities for collaboration between academia and industry.

More than just modelling and programming: educating engineers to leverage their digital skills in practice.

Modelling- and programming skills are essential for engineers. But while many study programs cover these technical skills, there is often limited attention for digital collaboration, data- and software management, and responsible and ethical application of modelling skills. With LLMs rapidly taking over menial coding tasks, the aspects become even more important.

To address this, our MSc course on modelling and computation replaced individual graded homework with a collaborative, staged, project-based approach. Student teams work on a single modelling project under the supervision of a teaching assistant acting as project owner (PO). The projects emulate an industrial setting: teams start with incomplete information, iteratively refine project scope and objectives with the PO, and build on previous results throughout the course. Project management and version control are conducted in GitLab.

Student performance is evaluated at both group and individual level through PO discussions and monitoring of individual Git contributions. Comprehension of AI-generated code is assessed through group–PO discussions. In this presentation, I will discuss the course design, implementation, and evaluation strategy, as well as lessons learned from three years of project evolution. Future developments, including a stronger focus on responsible AI/ML within the ERASMUS+ FAIR PLAI project, will also be discussed.

A phenomenological modeling approach to designing freezing processes for human induced pluripotent stem cell-derived spheroids

Human induced pluripotent stem (hiPS) cells form spontaneously cell aggregates called spheroids in cultivation. Spheroids have been applied to many clinical studies and applications. In the manufacturing, the freezing process is one of the most important steps because it is necessary for the transportation and preservation, although it has been difficult to freeze hiPS cell-derived spheroids. To improve the freezing technology, several studies were conducted using experimental [1] or computational modeling [2] approaches. However, the interaction of critical process parameters affecting the cell quality has still been unclear, resulting in limitation on rigorous process optimization.
　This work presents a phenomenological modeling approach to designing freezing processes for hiPS cell-derived spheroids. A model was developed that can consider the interaction of the cooling rate and the cryoprotective agent (CPA) concentration affecting the cell recovery ratio based on phenomenological modeling. The calculated recovery ratio was judged to be valid through freezing experiments. In the simulation using the developed model, the optimal parameter set of the cooling rate and the CPA concentration for a given size of the spheroids was obtained.

References
[1] Hiramatsu, et al., Journal of Parkinson's Disease (2022).
[2] Fujioka, et al., Digital Chemical Engineering (2026).

Cell-Tolerant Radial Affinity Chromatography for Capture of Monoclonal Antibodies from Unclarified Harvest

Cell-Tolerant Radial Affinity Chromatography (cTRAC) is a novel technology that uses packed beds of large-bead affinity resins that specifically bind a soluble product while allowing cells and debris to flow through, enabling direct capture from unclarified feed streams (Raedts, 2019). By integrating upstream and downstream processing, cTRAC has the potential to eliminate conventional clarification steps and thereby reduce product losses, product degradation, process time, footprint, and costs (Frej & Hjorth, 2018; Kinna et al., 2016).
This presentation will summarize results on the application of cTRAC to process unclarified mAb-containing CHO cell cultures, highlighting its feasibility as an integrated capture step and discussing key areas for further research and development.

References:
Frej, K. A.-K., & Hjorth, R. A. (2018). Expanded Bed Adsorption. In Biopharmaceutical Processing (pp. 269–277). Elsevier. https://doi.org/10.1016/B978-0-08-100623-8.00013-X
Kinna, A., Tolner, B., Rota, E. M., Titchener‐Hooker, N., Nesbeth, D., & Chester, K. (2016). IMAC capture of recombinant protein from unclarified mammalian cell feed streams. Biotechnology and Bioengineering, 113(1), 130–140. https://doi.org/10.1002/bit.25705
Raedts, M. J. H. (2019). Direct capture using large bead chromatography media (Patent No. NL2022427B1). https://patents.google.com/patent/NL2022427B1/en

CFD-DEM characterization of a 250 mL stirred-tank bioreactor for shear-sensitive stem-cell microcarrier cultures

In vitro cultivation of human inducible pluripotent stem cells (iPSC) for the production of Red Blood cells (RBC) is a promising therapeutic alternative to donor-based blood cell transfusions. The sensitivity of stem cell aggregates requires a tight balance between hydrodynamic stresses and suspension of the 3D aggregates, while maintaining sufficient oxygen transfer. We used CFD-DEM simulations to characterize a 250 mL bioreactor for shear sensitive microcarrier cultures. Our model uses Lattice Boltzmann large eddy simulation (LB-LES) to solve the fluid flow, combined with a discrete element method (DEM) for microcarrier particle motion.

We extensively validated our simulations using mixing time, kLa, and just-suspension experiments, demonstrating that our model can predict these engineering parameters with a 20% accuracy. Additionally, 4D-particle tracking velocimetry measurements showed a good agreement between measured and predicted particle velocity and acceleration profiles. The validated model was used to study the influence of important operating conditions, including stirrer speed, pumping-mode and filling volume. Furthermore, the Lagrangian nature of the DEM microcarriers was used to characterize the spatial-temporal fluctuations in energy dissipation rate from the microcarrier perspective to quantify hydrodynamic stresses. Future work will focus on applying this model to a 3L bioreactor and will help to scale-up of shear-sensitive stem-cell microcarrier cultures.

Engineering Simplified and Scalable Manufacturing Platforms for Circular RNA Therapeutics

RNA-based therapies are a powerful alternative to conventional vaccines due to their precision and adaptability. However, their broader implementation is limited by instability, delivery challenges, and the need for cost-effective, flexible manufacturing. Circular RNA (circRNA) addresses stability challenges through its covalently closed structure, lacking 5′ and 3′ ends and thus resisting exonuclease degradation. Despite these advantages, scalable and efficient production of circRNA remains a critical bottleneck.
Here, we present an integrated process engineering framework to simplify RNA-based therapy manufacturing. RNA circularisation is achieved via ribozyme-catalysed cleavage and ligation of transcripts in trans, enabling controllable, tunable and efficient conversion under mild conditions. Process analytics capability is enhanced through a novel RNA aptamer-based quantification method, allowing rapid and selective monitoring of circRNA formation. Downstream processing is intensified through sequential enzymatic cascades for targeted impurity degradation, significantly reducing reliance on chromatography-intensive purification trains. This approach lowers process complexity, facilitates scalability, and improves operational efficiency.
Overall, these simplified and modular approaches enable a more cost-effective and flexible manufacturing platform. This strategy also supports integration into continuous bioprocessing strategies improving yield and product quality.

Coffee break

PEGYLATION OF SILK FIBROIN NANOPARTICLES FOR DELIVERY OF 3,3’-DIINDOLYLMETHANE IN A ROUTE FOR OBESITY MANAGEMENT

Obesity is a major global health concern, and 3,3′-diindolylmethane (DIM) has shown promising anti-obesogenic activity. However, its poor stability and bioavailability limit its therapeutic application. This study developed PEGylated silk fibroin nanoparticles (PEG-SFNP) as a delivery platform for DIM. Silk fibroin nanoparticles (SFNP) were prepared by reverse microemulsion, incorporating DIM during synthesis for its loading. PEGylation was performed using methoxy PEG p-nitrophenyl carbonate (5 kDa). Drug loading and PEGylation efficiency were evaluated by HPLC. Nanoparticles were characterized by dynamic light scattering, ATR-FTIR spectroscopy, and scanning electron microscopy. Cytotoxicity was assessed in 3T3-L1 preadipocytes using the Alamar Blue assay. SFNP showed a drug loading capacity of 13.2 ± 4.7% for DIM and a PEGylation efficiency of 19.8 ± 6.3%. Nanoparticles were spherical, with sizes increasing from 85 ± 3 nm to 101 ± 2 nm after PEGylation, which also reduced the negative zeta potential from -36.1 ± 1.2 mV to -30.1 ± 1 mV. ATR-FTIR analysis confirmed successful DIM loading and PEG conjugation through changes in amide I and amide III bands, respectively. SFNP reduced 3T3-L1 cell viability to 71.4 ± 6% at 520 µg/mL, whereas PEG-SFNP showed no cytotoxicity. These results demonstrate that PEG-SFNP are a promising biocompatible platform for the delivery of anti-obesogenic compounds such as DIM and may improve their therapeutic potential for obesity management.

PROCESS DEVELOPMENT FOR SCALABLE PRODUCTION OF IPSC-DERIVED HEMATOPOIETIC STEM/PROGENITOR CELLS IN STIRRED-TANK BIOREACTORS

In vitro production of red blood cells (RBC) represents a promising therapy for treating inherited anemia patients. Using induced pluripotent stem cells (iPSC) for RBC production represent an unlimited cell source and have the potential to produce patient-specific cells.
A culture system for RBC generation was developed with minimal supporting cytokines, resulting in 40-60% enucleated erythroblasts. The 8-week protocol consists of 4 phases: 1) embryoid body (EB) formation, 2) hematopoietic organoid (HeO) formation using microcarriers (MC), 3) hematopoietic stem progenitor cells (HSPC) production, and 4) expansion of erythroblasts and terminal differentiation to RBC. We aimed to transfer the phase 2 to a stirred-tank bioreactor (STB) to produce iPSC -derived HSPC.
In this work, particle suspension, EB-MC attachment, and HeO yield were optimized by investigating the impact STB configuration, MC concentration, and intermittent stirring parameters.
The optimized bioreactor process resulted in HeO yield comparable to the dish control and reached a final cumulative production of 2.95∙108 HSPC. Additionally, the average volumetric productivity (6.8∙104 HSPC/mL/week) was more than 2-fold higher compared to the controls.
The optimized STB protocol demonstrates the reproducibility and feasibility of upscaling the production of iPSC-derived HSPC towards clinically relevant volumes in a fully controlled system, bringing us closer to manufacture cultured RBC transfusion units from iPSC.

Deep hybrid modelling of CHO platforms: addressing multiple development campaigns across different antibody products

Hybrid modelling that combines first-principles with machine learning is emerging as a cornerstone of Biopharma 4.0. Being the dominant host system in biopharma, CHO cultivation processes have been the object of several hybrid modelling studies. Most previous studies addressed single monoclonal antibody (mAb) models, combining macroscopic material balance equations with machine learning, predominantly shallow feedforward neural networks (FFNNs). Here, we address the challenge of developing hybrid models capable of handling multiple development campaigns across diverse mAb products, and investigate the implications for model architecture and selection. For this, a synthetic data set was produced by a process simulator for 72 antibodies by a DoE of 15 batches with varying feed rates and induction times. The adopted hybrid model structure combined a deep FFNN with mechanistic ODEs, with state feedback to the FFNN input layer (Figure 1). The training of the FFNN was coupled to the ODEs system using AdamW optimizer with stochastic regularization using mini-batch and weight dropout. An essential part of this work was model selection. We compared K-fold cross-validation with probabilistic metrics. Our results show that, regarding the prediction of a single mAb, a shallow hybrid structure is selected by both selection methods. However, for multiple mAb datasets, deep structures are required to ensure high predictive power.

Lunch break

Rotating Dynamic Crossflow Microfiltration as an Active Harvesting Platform for Continuous High-Cell-Density Escherichia coli Biomanufacturing

Continuous precision biomanufacturing requires harvesting technologies that remain controllable under changing solids loads. Classical harvest trains for high-cell-density Escherichia coli broths combine disc-stack centrifugation for biomass removal with filtration for residual solids. This two-step architecture is cost-intensive and can be challenged by feed variability.
Rotating dynamic crossflow microfiltration (DCF), adapted from food processing, is evaluated as an active harvesting platform. In DCF, a ceramic membrane rotates inside a closed chamber, generating shear directly at the membrane surface. This decouples shear generation from pump-driven bulk flow, addressing high-cell-density broths where solids loading and fouling control limit operation. Rotational speed becomes a handle to control cake formation.
Continuous Escherichia coli fermentation was integrated with DCF harvesting. Fresh high-cell-density broth was fed to DCF in chamber-retentive mode without continuous retentate outlet: clarified permeate was withdrawn continuously, while cells and solids were retained until discharge. Constant permeate flux was maintained for 48 h without sustained pressure increase, demonstrating stable harvesting with buffered solids loading.
By combining permeate removal, cell retention, hold-up and discharge in one membrane operation, DCF offers a route to consolidate harvest trains and turn harvesting into a controllable unit operation for model-based optimization.

Steric exclusion chromatography for scalable purification of ssDNA scaffolds in DNA-origami biomanufacturing

DNA-origami has emerged with great impact in nanotechnology due to its programmability. Its biomanufacturing relies on the production, purification and folding of a long single-stranded DNA (ssDNA) molecule, the scaffold. However, standard purification methods such as agarose gel extraction or ethanol precipitation are inefficient, cumbersome and not suitable for industrial applications.
In this work, steric exclusion chromatography (SXC) is explored as a novel strategy for the purification of ssDNA scaffolds directly from bacteriophage lysates. SXC relies on a mutual steric exclusion effect between polyethylene glycol (PEG) and a large target molecule in a hydrophilic stationary phase, enabling capture and purification without precipitate formation.
PEG precipitation was used to screen PEG conditions, where efficiency was found to be generally proportional to PEG size and concentration. SXC can be employed directly after phage expansion and lysis, and was tested with different PEG molecular weights (2000 to 8000 Da) and concentrations (10 to 20%). SXC enabled efficient recovery of ssDNA and effective impurity removal, outperforming standard purification methods. The purified scaffold was used to assemble a DNA-origami nanostructure, as confirmed by agarose gel electrophoresis and electron microscopy.
Overall, our results demonstrate the potential of SXC as an efficient and scalable downstream purification platform for ssDNA scaffold production in DNA-origami manufacturing.

Digital Twin aided Process Development and Enablement of Autonomous Manufacturing of Synthetic Peptides

Chromatographic separations are key unit operations in the downstream processing of biologics, including affinity and ion-exchange chromatography for capture, hydrophobic interaction and mixed-mode chromatography for intermediate purification, and reverse-phase chromatography for polishing. Mechanistic models such as the General Rate Model, the Steric Mass Action (SMA) isotherm, and lumped kinetic approaches have long underpinned model-based process development, forming the basis of Quality-by-Design strategies and Digital Twins, whose industrial relevance has further increased with ICH Q13.

Despite this maturity, robust parameter estimation from limited experimental data remains challenging. Nonlinear, highly correlated parameter spaces and strong sensitivity to initial guesses complicate reliable model calibration, particularly when transferring batch models to continuous processes.

This work presents a Python-based homotopy method for robust parameter estimation in mechanistic chromatography models. The approach gradually transforms a simplified surrogate problem into the target estimation problem, enabling stable convergence without proprietary software and facilitating integration into industrial Digital Twin workflows. The validated models are then used for in-silico design space exploration and batch-to-continuous process transfer, and are discussed in comparison with emerging approaches such as PINNs and Bayesian optimization.

A Mechanistic Model for Predicting AAV9 Recovery During Ion-Exchange Purification

Ion-exchange chromatography (IEX) is widely used for AAV purification and separation of genome-containing capsids from product-related impurities due to its scalability and robustness. However, purification remains challenging because AAV preparations contain heterogeneous populations of full, empty, and partially filled capsids, each exhibiting significant charge variability caused by structural differences and post-translational modifications. Consequently, these populations behave as continuous charge distributions rather than discrete species, resulting in broad overlapping elution profiles and co-elution during IEX. Here, we present a mechanistic framework that explicitly accounts for charge heterogeneity within AAV capsid populations. The model was parameterized and validated using extensive experimental data from linear-gradient, two-step, and multistep IEX elution strategies across varying pH, flow rates, loading conditions, and feed compositions. The framework accurately predicts chromatographic profiles, fraction-specific capsid compositions, and recovery yields, while also capturing conductivity-dependent loss phenomena. Model-based analyses reveal intrinsic trade-offs between purity and recovery caused by the broad charge distributions of capsid subpopulations. Overall, the framework provides a quantitative tool for rational optimization of IEX purification processes and can be extended beyond AAV9 to other serotypes with heterogeneous chromatographic behavior.

Coffee break

Dairy Side-Stream Valorization Through Microbial Lipid Production: Comparative Assessment of Yarrowia lipolytica and Candida kefyr

Dairy waste streams are lactose-rich substrates with potential for microbial lipid production but lactose utilization differs among oleaginous yeasts. This study evaluated a culture collection strain of Yarrowia lipolytica and an indigenous dairy-related Candida kefyr isolate for growth and lipid production on whey. Whey powder contained 68% lactose, 11% protein, and 2% lipids at pH 6. The oleaginous nature of both strains was confirmed in a glucose-based medium, yielding biomass concentrations of 3.46 and 2.71 g/L and lipid contents of 29% and 20% for Y. lipolytica and C. kefyr, respectively. Lactose assimilation tests on agar containing 5-20 g/L lactose showed that C. kefyr grew across all tested range, whereas Y. lipolytica grew only up to 10 g/L. In liquid cultures with 5 g/L lactose, both strains grew efficiently; however, at 20 g/L lactose, C. kefyr increased from 10² to 10⁷ CFU/mL, while Y. lipolytica was strongly inhibited. Cultivation of C. kefyr on non-optimized whey medium containing 20 g/L lactose produced 3 g/L biomass and lipid contents exceeding 30% of dry cell weight. Medium optimization identified a C/N ratio of 68 as favorable for lipid accumulation and will be used in subsequent optimal whey-cultivations. These results highlight the importance and potential of indigenous yeasts for the conversion of lactose-rich side streams into microbial lipids.
Keywords: Oleaginous yeasts; Whey valorization; Lactose utilization; Microbial lipids; Circular bioeconomy

Real-time monitoring of stress responses of lactic acid bacteria for industrial applications

Lactic acid bacteria (LAB) are increasingly being used in more complex cultivations aside established batch processes, yet suitable monitoring tools for assessing cell physiology are lacking. Since it is assumed that LAB respond to environmental stress with an altered detachment behavior during binary fission, this can lead to increased cell length and the formation of cell chains [1]. A combined approach involving on-line volumetric determination of cell and chain length using a novel optical stop-flow measurement technique and at-line microscopy together with automated image recognition was applied in cultures of Lactobacillus paracasei DSM6433 and Lactococcus lactis DSM4366 to demonstrate that stressful cultivation conditions can be reliably monitored by combining the two methods. They were applied in axenic batch and fed-batch cultures, as well as in co-cultures with yeast.
It was demonstrated that the consideration of cell morphology at the single-cell level and the rapid determination of average cell length serve as early indicators of reduced growth and metabolic turnover. This relevant correlation of stress responses leads to the identification of suitable growth conditions across various scales and enables new processes for the industrial use of LAB.

Acknowledgements
The authors kindly acknowledge financial support by Novo Nordisk Fonden, grant no. NNF22OC0081345.

[1] Clausen, A. K., Junne, S. (2025), World J Microbiol Biotechnol, DOI:10.1007/s11274-025-04385-9

Yarrowia-driven solid-state fermentation of Portuguese fruit surpluses

The transition towards sustainable food systems has intensified the search for innovative strategies to valorize agricultural surpluses. In Portugal, the production of regionally certified pome fruits, such as Gala and Fuji apples (PGI) and “Rocha” pears (DOP), generates large amounts of fruits rejected by the retail market due to quality and appearance standards, representing an important environmental and economic challenge. Thus, this work proposes an integrated bioprocess to valorize apple and pear surpluses by enhancing their protein content through solid-state fermentation (SSF) with Yarrowia spp.
A screening of Yarrowia strains was conducted under submerged fermentation to evaluate protein-rich biomass production using fructose as a carbon source. Yarrowia lipolytica W29 and Yarrowia divulgata 5257/2 were the most promising strains for SSF with apple and pear surpluses producing the highest yeast biomass and protein. SSF led to 5-fold increase in protein and 38% reduction in sugars, yielding a high-protein, low-sugar fermented substrate suitable as food ingredients. Through a Plackett–Burman design of experiments, moisture content and ammonium sulfate supplementation were the main factors influencing biomass and protein production in SSF with both fruits.
This SSF strategy highlights the potential to convert fruit surpluses into protein-rich ingredients through microbial fermentation, supporting circular economy and creating opportunities for regional producers.

Life cycle assessment of integral production chains for cellular agriculture products

The sustainability benefits of cellular agriculture products, namely cultured meat (CM) and alternative microbial proteins, can be ascertained by means of Life Cycle Assessments (LCA). These are crucial tools to aid decision making in process design, as well as to support entrepreneurs and policy makers in the commercialization and regulation of these novel foods. Previous LCA studies in the field have been conducted at very early stages of development, using extrapolated scientific data and observations at small-scale to predict the environmental performance of future facilities. As a result, the lack of harmony concerning system boundaries, type of data, and assumptions used in such studies, leads to high variability on environmental impact results.
Such studies also fail to capture the impact of novel process configurations and most recent findings in the field, namely data resulting from current upscaling efforts by industrial partners at the Cellular Agriculture Netherlands (CAN).
This works investigates the environmental impact and establishes a comprehensive Life Cycle Inventory (LCI) concerning integral production chains for cellular agriculture meat and dairy products, from raw materials to final disposal, considering two different scenarios for technology development: 1) current state-of-the-art, and 2) future projections, based on data generated by prior and ongoing research within the core research program of CAN.

Lunch break

A Platform Approach to Cultivated Meat: scalable cells, high-throughput screening, and computational modeling

The development of robust cell culture platforms for cultivated meat production requires careful consideration of intrinsic cellular characteristics and process compatibility. In this work, we outline key requirements for a platform cell line for cultivated meat. This includes immortal growth capacity to enable extensive expansion, adaptation to serum- and animal component–free media for improved ethical and economic feasibility, and the ability to grow in single-cell suspension to facilitate straightforward scale-up. While alternative cell lines and modalities may be suitable for specific applications, the selected characteristics provide a broadly compatible foundation across diverse bioprocess configurations.
A fibroblast-like quail cell line fitting these characteristics is demonstrated to have consistent growth performance across multiple laboratory scales, including deep multi-well plates, shake flasks, and mini-bioreactor systems, supporting its application in both early-stage screening and process development. Complementing this, a high-throughput cell-viability assay has been established to enable evaluation of culture conditions on animal single-cell suspension cell lines. The assay is utilized to determine key kinetic parameters, such as inhibition profiles of waste-metabolites, and medium design- and optimization studies. Experimental data is used to create mechanistic cell culture simulations, providing insight into technical and economic performance at scale.

Cultured seafood: a tissue engineering approach

The development of structured seafood products may foster consumers acceptance. While obtaining high yields of cell biomass at scale, with high nutritional value, low environmental impacts and affordable costs is crucial; food manufacture methodologies become available to deliver premium products with the right texture and taste. This presentation lays strategies on structuring and texturing seafood. Results presented highlight advances made at our laboratory, including strategies used on fabrication of cultured fish fillet combining 3D bioprinting and electrospinning techniques. Specifically, zein-based and wheat-based nanofibers were produced and combined with 3D bioprinting of novel myogenic and fat bioinks. Plant sourced proteins, carrageenan, alginate, and methylcellulose were combined to formulate bioinks used for fish cells 3D bioprinting. Finally, the 3D bioprinted cultured fish fillet organoleptic properties were enhanced by adding microalgae biomass and omega-3.

The authors acknowledge funding from Fundação para a Ciência e Tecnologia and FEDER (Lisboa2030), for the PhD scholarships (2022.12900.BD, 2023.00599.BD), CleanFish (2022.07677.PTDC)), Mack4All (14944-LISBOA2030-FEDER-00855900-2023.17995.ICDT), iBB (UIDB/04565/2025 and UIDP/04565/2025), i4HB (LA/P/0140/2020); and European Commission through FEASTS (ID 101136749) and Unite! Widening (PhD Fellowship Unite! ULisboa 13BD2025). Views and opinions expressed of the authors only and not of the European Union.

How large can you go? Model-based screening of bioreactors for large scale cultivated meat production.

Scaling up bioprocesses for cultivated meat remains a major technological bottleneck, driven by uncertain requirements for mass transfer, shear tolerance, and mixing performance. We developed a two-stage evaluation approach for conceptual reactor designs. This uses a Monte Carlo–based timescale assessment for initial global screening of feasible stirred tank (STR), bubble column (BC) and airlift reactor (ALR) designs, followed by compartmental modelling to quantify possible gradient formation. Designs were filtered against timescales for mixing, O₂ transfer and CO₂ accumulation while imposing a bubble shear constraint via the Hypothetical Killing Volume (HKV). The analysis identified mixing and O₂ transfer as dominant constraints and showed that geometric optimization and shear protectants can expand the feasible design window.

Guided by these insights, we developed an industrial scale 1D compartment model for ALRs up to 250 m³. The model resolves height dependent gas hold up, circulation velocities, and local kLa, yielding liquid and gas phase O₂ and CO₂ profiles. The model quantifies starvation and toxicity thresholds that emerge along the reactor height and shows feasible biomass concentrations depend strongly on geometric parameters and cell characteristics. Together, the framework provides a basis for ALR scale up and actionable guidance on cell engineering and operating windows for cultivated meat production.

Integrating upstream cultivations and downstream processes: The impact of cultivation conditions on intracellular protein release from mildly disrupted Saccharomyces cerevisiae